Spin delocalization in electron-rich iron(III) piano-stool σ-acetylides. An experimental (NMR) and theoretical (DFT) investigation

874

Organometallics 2007, 26, 874-896

Spin Delocalization in Electron-Rich Iron(III) Piano-Stool

σ-Acetylides. An Experimental (NMR) and Theoretical (DFT)

Investigation

Fre´de´ric Paul,*^,†Gre´gory da Costa,^‡Arnaud Bondon,*^,‡Nicolas Gauthier,^†

Sourisak Sinbandhit,^§Loic Toupet, Karine Costuas,^†Jean-Franc¸ois Halet,^†and

Claude Lapinte^†

Sciences Chimiques de Rennes, UMR 6226 CNRS, UniVersite´ de Rennes 1, Campus de Beaulieu, 35042

Rennes Cedex, France, RMN-ILP, UMR 6026 CNRS, UniVersite´ de Rennes 1, IFR 140-PRISM, CS 34317,

Campus de Villejean, 35043 Rennes Cedex, France, CRMPO, UniVersite´ de Rennes 1, Campus de

Beaulieu, 35042 Rennes Cedex, France, and Groupe Matie`re Condense´e et Mate´riaux, UMR 6626 CNRS,

UniVersite´ de Rennes I, Campus de Beaulieu, 35042 Rennes Cedex, France

ReceiVed October 26, 2006

Several paramagnetic electron-rich Fe(III) mononuclear arylacetylide complexes of formula [(η²-

dppe)(η⁵-C₅Me₅)Fe(CtC-Ar)]⁺in which Ar represents a functional aryl group were studied by means

of multinuclear NMR. All signals detected for the various nuclei were assigned. Hyperfine coupling

constants for selected nuclei of the arylacetylide ligand were derived from ¹H or ¹⁹F NMR contact shifts.

These NMR data are diagnostic of a metal-centered unpaired electron partly residing in a π molecular

1

orbital on the arylacetylide ligand, in line with DFT computations. We show here that the H NMR

paramagnetic shifts of the ortho (H₁) and meta (H₂) arylacetylide protons convey decisive information

on the charge distribution in the aryl ring. Estimates of the relaxation rates of the unpaired electron were

1

also derived from half-widths of the H NMR signals. Finally, line-broadening studies of Fe(II)/Fe(III)

mixtures allowed extracting the self-exchange rates for several redox couples among these complexes.

The self-exchange rates appear slightly substituent dependent and are apparently larger for compounds

with electron-withdrawing substituents on the aryl ring. Reorganization energies of ca. 4000 cm^-1could

be derived for these outer-sphere electron-transfer processes.

switchable components,^2-4not only as single molecule devices

but also as molecular precursors for attaining new materials with

specifically tailored properties.^3f,5A constant feature usually

observed with such compounds is the strong dependence of their

electronic structure and properties on the redox state of the

appended metal center.^3a,6,7Thus, in suitably designed carbon-

rich compounds, it has been shown in several instances that

the overall enhancement (or inhibition) of a given electronic

Introduction

These last decades, in the emerging field of molecular

electronics,¹mono- or polynuclear organometallic complexes

featuring carbon-rich ligands and stable over several redox states

have aroused a lot of academic interest as potential redox-

* Corresponding authors. Tel: 33 (0)2 23 23 59 62. Fax: 33 (0)2 23 23

56 37. E-mail: frederic.paul@univ-rennes1.fr. Tel: 33 (0)2 23 23 65 61.

Fax: 33 (0)2 23 23 46 06. E-mail: arnaud.bondon@univ-rennes1.fr.

^†UMR CNRS 6226: Sciences Chimiques de Rennes.

(3) (a) Ren, T. Organometallics 2005, 24, 4854-4870. (b) Powell, C.

E.; Humphrey, M. G. Coord. Chem. ReV. 2004, 248, 725-756. (c) Rigaut,

S.; Touchard, D.; Dixneuf, P. H. Coord. Chem. ReV. 2004, 248, 1585-

1601. (d) Bruce, M. I.; Low, P. J. AdV. Organomet. Chem. 2004, 50, 179-

444. (e) Ceccon, A.; Santi, S.; Orian, L.; Bisello, A. Coord. Chem. ReV.

2004, 248, 683-724. (f) Long, N. J.; Williams, C. K. Angew. Chem., Int.

Ed. 2003, 42, 2586-2617. (g) Yam, V. W.-W. Acc. Chem. Res. 2002, 35,

555-563.

^‡UMR CNRS 6026: RMN-ILP.

^§CRMPO.

UMR CNRS 6626: Groupe Matie`re Condense´e et Mate´riaux.

(1) (a) Robertson, N.; Mc. Gowan, G. A. Chem. Soc. ReV. 2003, 32,

96-103. (b) Caroll, R. L.; Gorman, C. B. Angew. Chem., Int. Ed. Engl.

2002, 41, 4379-4400. (c) Tour, J. M. Acc. Chem. Res. 2000, 33, 791-

803.

(2) (a) Gao, L.-B.; Liu, S.-H.; Zhang, L.-Y.; Shi, L.-X.; Chen, Z.-N.

Organometallics 2006, 25, 506-512. (b) Gao, L.-B.; Zhang, L.-Y.; Shi,

L.-X.; Chen, Z.-N. Organometallics 2005, 24, 1678-1684. (c) Qi, H.;

Ghupta, A.; Noll, B. C.; Snider, G. L.; Lu, Y.; Lent, C. S.; Fehlner, T. P.

J. Am. Chem. Soc. 2005, 127, 15218-15227. (d) Blum, A. S.; Ren, T.;

Parish, D. A.; Trammell, S. A.; Moore, M. H.; J. G. Kushmerick; Xu, G.-

L.; Deschamps, J. R.; Polack, S. K.; Shashidar, R. J. Am. Chem. Soc. 2005,

127, 10010-10011. (e) Wong, W.-Y.; Che, C.-M.; Chan, M. C. W.; Han,

J.; Leung, K.-H.; Phillips, D. L.; Wong, K.-Y.; Zhu, N. J. Am. Chem. Soc.

2005, 127, 13997-14007. (f) Xu, G.-L.; Crutchley, R. J.; DeRosa, M. C.;

Pan, Q.-J.; Zhang, H.-X.; Wang, X.; Ren, T. J. Am. Chem. Soc. 2005, 127,

13354-13365. (g) Hu, Q. Y.; Lu, W. X.; Tang, H. D.; Sung, H. H. Y.;

Wen, T. B.; Williams, I. D.; Wong, G. K. L.; Lin, Z.; Jia, G. Organome-

tallics 2005, 24, 3966-3973. (h) Venkatesan, K.; Blacque, O.; Fox, T.;

Alfonso, M.; Schmalle, H. W.; Berke, H. Organometallics 2004, 23, 1183-

1186. (i) Fraysse, S.; Coudret, C.; Launay, J.-P. J. Am. Chem. Soc. 2003,

125, 5880-5888. (j) Naklicki, M. L.; White, C. A.; Plante, L. L.; Evans,

C. E. B.; Crutchley, R. J. Inorg. Chem. 1998, 37, 1880-1885.

(4) (a) Cifuentes, M. P.; Humphrey, M. G.; Morall, J. P.; Samoc, M.;

Paul, F.; Roisnel, T.; Lapinte, C. Organometallics 2005, 24, 4280-4288.

(b) Hurst, S. K.; Cifuentes, M. P.; Morrall, J. P. L.; Lucas, N. T.; Whittal,

I. R.; Humphrey, M. G.; Asselberghs, I.; Persoons, A.; Samoc, M.; Luther-

Davies, B.; Willis, A. C. Organometallics 2001, 20, 4664-4675. (c) Powell,

C. E.; Humphrey, M. G.; Cifuentes, M. P.; Morall, J. P.; Samoc, M.; Luther-

Davies, B. J. Phys. Chem. A 2003, 107, 11264-11266.

(5) (a) Schwab, P. F. H.; Smith, J. R.; Michl, J. Chem. ReV. 2005, 105,

1197-1279. (b) Schwab, P. F. H.; Levin, M. D.; Michl, J. Chem. ReV.

1999, 99, 1863-1933. (c) Fillaut, J.-L.; Perruchon, J.; Blanchard, P.;

Roncali, J.; Gohlen, S.; Allain, M.; Migalska-Zalas, A.; Kityk, I. V.;

Sahraoui, B. Organometallics 2005, 24, 687-695. (d) Szafert, S.; Gladysz,

J. A. Chem. ReV. 2003, 103, 4175-4205.

(6) (a) Paul, F.; Lapinte, C. Coord. Chem. ReV. 1998, 178/180, 427-

505. (b) Paul, F.; Lapinte, C. In Unusual Structures and Physical Properties

in Organometallic Chemistry; Gielen, M., Willem, R., Wrackmeyer, B.,

Eds.; Wiley & Sons, Ldt: San-Francisco, 2002; pp 219-295.

(7) Crutchley, R. J. AdV. Inorg. Chem. 1994, 41, 273-325.

Publication on Web 01/17/2007

Spin Delocalization in Electron-Rich Iron(III) σ-Acetylides

Organometallics, Vol. 26, No. 4, 2007 875

Chart 1

property such as fluorescence or nonlinear optical (NLO) activity

can be achieved upon variation of the redox state of the metal

center.^4,8-9

j[PF₆] (Chart 1) were dominantly metal-centered S ) 1/2 radical

cations.⁶In addition, DFT calculations on computationally

simpler models have suggested that significant delocalization

of the electronic vacancy takes place on the arylacetylide ligand

in compounds featuring strongly electron-releasing substitu-

ents.¹⁵If experimentally confirmed, this theoretical prediction

constitutes a very important result for better understanding the

unexpectedly large through-bridge exchange coupling sometimes

observed in related polynuclear Fe(III) assemblies.^6b,10c,16

Unfortunately, no direct experimental confirmation of these

theoretical predictions could be obtained to date. Indeed, ESR

measurements, which usually provide estimates of the spin

delocalization in a straightforward way, proved useless in that

respect, since no (super)hyperfine structures were observed on

the spectra recorded with frozen solutions of 1a-j[PF₆], due

to the fast electronic relaxation of these Fe(III) samples.¹⁵

For several years now, our group has focused on electron-

rich iron piano-stool acetylide complexes featuring “(η²-

dppe)(η⁵-C₅Me₅)Fe(CtC)-” fragments (dppe ) 1,2-bis-

(diphenylphosphino)ethane).¹⁰Polynuclear architectures of this

kind have revealed an appealing potential to realize molecular-

scaled wires or magnets,^6,10and also redox-switchable photonic

devices, especially when associated with arylethynyl ligands.^4a,8

Now, a prerequisite to design more efficient molecular as-

semblies possessing similar properties is to obtain a good

understanding of the electronic perturbation induced by the iron

center on the nearby carbon-rich ligand, depending on its redox

state. To date, this end-group, as well as its Ru(II) analogue,¹¹

was conclusively demonstrated to behave as an electron-

releasing group resembling a methoxy or amino substituent in

the closed-shell Fe(II) state.^12-14Its electronic behavior is

however less well defined in the open-shell Fe(III) state.¹⁵Up

to now, UV-visible-NIR and ESR investigations have clearly

shown that mononuclear Fe(III) model complexes such as 1a-

Under such circumstances, in favorable cases, it is well known

that NMR spectroscopy can allow determination of hyperfine

couplings, providing not only their magnitude but also their

sign.^17,18We therefore decided to initiate a NMR study of the

known paramagnetic compounds 1a-j[PF₆] and 2[PF₆], and also

of the new paramagnetic 3a,b[PF₆] complexes in which the

arylacetylide has been “tagged” with methyl groups. Our first

objective was to assign all the observable signals and, if possible,

to empirically map the spin delocalization on the arylacetylide

ligand in 1a-j[PF₆] depending on the X substituent’s nature.

A second objective was then to compare experimental (NMR)

with theoretical (DFT) estimates of the spin delocalization in

these compounds, in order to determine the consistency of DFT

calculations with experiment.

(8) (a) Wong, K. M.-C.; Lam, S. C.-F.; Ko, C.-C.; Zhu, N.; Yam, V.

W.-W.; Roue´, S.; Lapinte, C.; Fathallah, S.; Costuas, K.; Kahlal, S.; Halet,

J.-F. Inorg. Chem. 2003, 42, 7086-7097. (b) Paul, F.; Costuas, K.; Ledoux,

I.; Deveau, S.; Zyss, J.; Halet, J.-F.; Lapinte, C. Organometallics 2002, 21,

5229-5235. (c) Weyland, T.; Ledoux, I.; Brasselet, S.; Zyss, J.; Lapinte,

C. Organometallics 2000, 19, 5235-5237.

(9) For an inorganic example, see also: Coe, B. J.; Houbrechts, S.;

Asselberghs, I.; Persoons, A. Angew. Chem., Int. Ed. 1999, 38, 366-368.

(10) (a) Ibn Ghazala, S.; Paul, F.; Toupet, L.; Roisnel, T.; Hapiot, P.;

Lapinte, C. J. Am. Chem. Soc. 2006, 128, 2463-2476. (b) de Montigny,

F.; Argouarch, G.; Costuas, K.; Halet, J.-F.; Roisnel, T.; Toupet, L.; Lapinte,

C. Organometallics 2005, 24, 4558-4572. (c) Bruce, M. I.; Low, P. J.;

Hartl, F.; Humphrey, P. A.; de Montigny, F.; Jevric, M.; Lapinte, C.; G. J.

Perki, S.; Roberts, R. L.; Skelton, B. W.; White, A. H. Organometallics

2005, 24, 5241-5255. (d) Paul, F.; Meyer, W.; Jiao, H.; Toupet, L.; Gladysz,

J. A.; Lapinte, C. J. Am. Chem. Soc. 2000, 122, 9405-9414. (e) Le Stang,

S.; Paul, F.; Lapinte, C. Organometallics 2000, 19, 1035-1043. (f) Le

Narvor, N.; Toupet, L.; Lapinte, C. J. Am. Chem. Soc. 1995, 117, 7129-

7138.

Much to our surprise, while comparing our results with

previous investigations, we realized that although ¹H NMR had

often been used as a convenient experimental tool to reveal or

study (para)magnetism in related mono- or polynuclear orga-

nometallic complexes with carbon-rich ligands,^10,19-23only

(11) Paul, F.; Ellis, B. J.; Bruce, M. I.; Toupet, L.; Roisnel, T.; Costuas,

K.; Halet, J.-F.; Lapinte, C. Organometallics 2006, 25, 649-665.

(12) Costuas, K.; Paul, F.; Toupet, L.; Halet, J.-F.; Lapinte, C. Orga-

nometallics 2004, 23, 2053-2068.

(16) Roue´, S.; Le Stang, S.; Toupet, L.; Lapinte, C. C. R. Chim. 2003,

6, 353-366.

(17) Bertini, I.; Luchinat, C. NMR of Paramagnetic Molecules in

Biological Systems; The Benjamin/Cummings Publishing Co.: Menlo Park,

CA, 1986.

(13) Paul, F.; Mevellec, J.-Y.; Lapinte, C. J. Chem. Soc., Dalton Trans.

2002, 1783-1790.

(18) McConnell, H. M.; Chesnut, D. B. J. Chem. Phys. 1958, 28, 107-

117.

(19) Bruce, M.; Costuas, K.; Davin, T.; Ellis, B. J.; Halet, J.-F.; Lapinte,

C.; Low, P. J.; Smith, M. E.; Skelton, B. W.; Toupet, L.; White, A. H.

Organometallics 2005, 24, 3864-3881.

(14) Denis, R.; Toupet, L.; Paul, F.; Lapinte, C. Organometallics 2000,

19, 4240-4251.

(15) Paul, F.; Toupet, L.; The´pot, J.-Y.; Costuas, K.; Halet, J.-F.; Lapinte,

C. Organometallics 2005, 24, 5464-5478.

876 Organometallics, Vol. 26, No. 4, 2007

Scheme 1. (a) Ar ) 3,5-Xyl and (b) Ar ) 2,4,5-Mes

Paul et al.

seldomly was a complete assignment of the observed signals

provided.^21,23-26More specifically with mononuclear acetylide

complexes, a complete multinuclear NMR study aimed at

assessing the importance of spin delocalization was conducted

only once, to our knowledge,²⁶and not with electron-rich aryl

acetylides. This prompted us to communicate the present

classical syntheses (Scheme 1). All the new Fe(II) and Fe(III)

complexes were fully characterized and presented the expected

spectroscopic features (Experimental Section).

We report in Figure 1 (a and b) the solid-state structures of

1c[PF₆] and m-1d[PF₆], which were obtained by slow diffusion

of n-pentane in a CH₂Cl₂solution of the corresponding

complexes. The former is a known compound that was fully

characterized previously,¹⁵while m-1d[PF₆] is the meta-

analogue of the known 1d[PF₆].²⁸There is nothing exceptional

about distances and angles or the packing of these two

complexes, but accurate geometrical parameters were needed

for NMR calculations (see hereafter). The structures of the

complexes 3a, 3b, and 3b[PF₆] in the solid state were also

confirmed by X-rays (Figure 1, parts c and d). The packing in

the solid state as well as bond distances and angles are usual

for these Fe(II) and Fe(III) arylacetylide complexes (Experi-

mental Section), except perhaps for the torsion angle in the range

120-140° adopted by the phenyl plane with respect to the

metal-C₅Me₅axis in 3a, 3a[PF₆], and 3b[PF₆].^12,14,15,28Indeed,

angles closer to 90° were most often observed in related Fe(II)

and Fe(III) complexes.^12,15In line with the additional steric bulk

induced by the methyl groups on the arylalkynyl ligands, these

conformations allow minimizing the intramolecular repulsive

interactions with the dppe-aryl groups and the C₅Me₅-methyl

groups in the solid state. In contrast, the strong bending of the

C37-C38-C39 axis (171.9°) observed in 3a cannot originate

from intramolecular interactions, since such a feature is not

observed for 3b, presenting an even more hindered phenyla-

lkynyl ligand. It certainly finds its origin in packing forces, as

already observed for related compounds.^12,14

1

contribution, which usefully complements previous H NMR

studies made on 2[PF₆] and 1f[PF₆] (Chart 1).^21,22Before

discussing the NMR results, we will start by (i) briefly giving

the synthesis and characterization of the new complexes 3a[PF₆]

1

and 3b[PF₆]. We will next (ii) report and assign the H, ¹³C,

³¹P, and ¹⁹F (for fluorine-containing compounds) NMR spectra

of 1a-i[PF₆], 3a,b[PF₆], and some related known Fe(III)

complexes, before (iii) deriving the proton and fluorine hyperfine

coupling constants as well as (iv) the spin densities on selected

carbon atoms of the aryl ring of these compounds. (v) The

temperature dependence and (vi) the half-widths of the various

signals will also be briefly examined, before (vii) deriving the

self-exchange rates for selected complexes using line-broadening

studies. Finally, (viii) the spin delocalization taking place in

these Fe(III) acetylide compounds will be analyzed on more

theoretical grounds with the help of DFT calculations.

Results

Synthesis and Characterization of the New Complexes 3a-

[PF₆] and 3b[PF₆]. The known cationic Fe(III) complexes 1a-

j[PF₆] and 2[PF₆] were obtained as previously described.^14,22,27

The new Fe(III) complexes m-1d[PF₆], 3a[PF₆], and 3b[PF₆]

were isolated in a similar way from the Fe(II) parents m-1d,

3a, and 3b by oxidation using ferricinium hexafluorophosphate.

The new Fe(II) acetylides 3a and 3b were themselves obtained

from the Fe(II) chloride complex 2 and the preformed terminal

alkynes 6a,b via the vinylidene complexes 7a,b[PF₆] using

Assignment of the ¹H NMR Spectra. The NMR spectra of

the various samples have been recorded in either dichlo-

romethane-d₂or deuterated chloroform. While similar results

were obtained in both solvents, the former was usually preferred

for most investigations, since a slow reaction of the Fe(III)

complexes was observed in the latter,¹⁵presumably generating

the corresponding vinylidene complexes by hydrogen atom

abstraction,^10f,11even under strict absence of oxygen.²⁹

In this study, the known chloride complex 2[PF₆] was used

as a benchmark for identifying the signals pertaining to the

“(η²-dppe)(η⁵-C₅Me₅)Fe” core. We therefore started our inves-

tigation by monitoring the shifts of various mixtures of Fe(II)

and Fe(III) redox congeners with 1a/1a[PF₆], 1g/1g[PF₆], and

2/2[PF₆]. For all these compounds, the electron self-exchange

reaction proved to be much faster than the ¹H NMR time scale

under the measurement conditions used (see below), and an

(20) (a) Kheradmandan, S.; Venkatesan, K.; Blacque, O.; Schmalle, H.;

Berke, H. Chem.-Eur. J. 2004, 10, 4872-4885. (b) Fernandez, F. J.;

Blacque, O.; Alfonso, M.; Berke, H. Chem. Commun. 2001, 1266-1267.

(21) Weyland, T.; Costuas, K.; Mari, A.; Halet, J.-F.; Lapinte, C.

Organometallics 1998, 17, 5569-5579.

(22) Roger, C.; Hamon, P.; Toupet, L.; Rabaaˆ, H.; Saillard, J.-Y.; Hamon,

J.-R.; Lapinte, C. Organometallics 1991, 10, 1045-1054.

(23) Ko¨hler, F. H.; Pro¨ssdorf, W.; Schubert, U. Inorg. Chem. 1981, 20,

4096-4101.

(24) (a) Venkatesan, K.; Fox, T.; Schmalle, H. W.; Berke, H. Organo-

metallics 2005, 24, 2834-2847. (b) Krivyk, V. V.; Eremenko, I. L.; Veghini,

D.; Petrunenko, I. A.; Poutney, D. L.; Unseld, D.; Berke, H. J. Organomet.

Chem. 1996, 511, 111-114.

(25) Balch, A. L.; Latos-Grazynski, L.; Noll, B. C.; Philips, S. L. Inorg.

Chem. 1993, 32, 1124-1129.

(26) Ko¨hler, F. H.; Hofmann, P.; Pro¨ssdorf, W. J. Am. Chem. Soc. 1981,

103, 6359-6372.

(28) Courmarcel, J.; Le Gland, G.; Toupet, L.; Paul, F.; Lapinte, C. J.

Organomet. Chem. 2003, 670, 108-122.

(29) Paul, F.; Toupet, L.; Roisnel, T.; Hamon, P.; Lapinte, C. C. R. Chim.

2005, 8, 1174-1185.

(27) Tilset, M.; Fjeldahl, I.; Hamon, J.-R.; Hamon, P.; Toupet, L.;

Saillard, J.-Y.; Costuas, K.; Haynes, A. J. Am. Chem. Soc. 2001, 123, 9984-

10000.

Spin Delocalization in Electron-Rich Iron(III) σ-Acetylides

Organometallics, Vol. 26, No. 4, 2007 877

Figure 1. ORTEP representations of the cation of 1c[PF₆] (a), the cation of m-1d[PF₆] (b), the compound 3a (c), and the cation of 3b[PF₆]

(d) at the 50% probability level.

1

Figure 2. Observed H NMR shifts for 1a/1a[PF₆] mixtures in CD₂Cl₂at 25 °C (assignment according to Chart 2). Aromatic protons of

the dppe ligand are shown on the diagram at the right (b). For all fits R²> 0.99.

averaged single set of signals was always obtained.²²Progressive

increase in the Fe(III) concentration allowed identifying most

of the signals detected for the paramagnetic complexes as shown

with 1a[PF₆] in CD₂Cl₂(Figure 2). Similar diagrams obtained

for the para-tolyl alkynyl complex 1g[PF₆] in CD₂Cl₂and for

the chloride complex 2[PF₆] in CDCl₃are provided as Sup-

porting Information. From these Fe(II)/Fe(III) correlations,

complemented by a combination of COSY and NOESY

experiments on both the diamagnetic Fe(II) and paramagnetic

Fe(III) compounds 1a-k/1a-k⁺and 3a,b/3a,b⁺, a definitive

assignment (Table 1) could be gained for all the protons of the

various complexes (Supporting Information). As an example,

the typical spectrum of 1a[PF₆] is shown in Figure 3, along

with the corresponding assignment of the protons (Chart 2).

In all compounds, rather specific shifts were observed for

the “(η²-dppe)(η⁵-C₅Me₅)Fe” core, identifying the peaks belong-

ing to H₁and H₂in a straightforward way (Figure 3). These

signals are usually the most upfield- and downfield-shifted peaks

on the spectra of the various compounds. For 3a[PF₆] and

3b[PF₆], which both present methyl groups in place of these

hydrogen atoms (Chart 1), the low-field and high-field signals

are respectively lacking, confirming that the former signal

corresponds to the meta hydrogen atoms of the arylacetylide

ligand, while the latter corresponds to the ortho ones. In addition,

new signals were detected at high field (-13.3 ppm) for the

meta methyl groups of 3a[PF₆] and at low field (47.4 ppm) for

the ortho methyl groups for 3b[PF₆]. The para hydrogen of the

arylacetylide ligand of 3a[PF₆] is detected at high field (-46.5

878 Organometallics, Vol. 26, No. 4, 2007

Paul et al.

Table 1. ¹H NMR Shifts (δ (0.1 ppm) Recorded for the [(η²-dppe)(η⁵-C₅Me₅)FeR][PF₆] Complexes at 20 °C in CD₂Cl₂

a

compd

R

-Ct C-4-(C₆H₄)-

dppe

H₅

C₅Me₅

H₁₀

X

H₁

H₂

H₃

H₄

H₆

2.9

H₇

H₈

H₉

H₁₁

H_X

Cl (2⁺)

7.6

5.3

5.5

5.9

6.6

6.8

7.2^b

7.0

6.8

6.7

6.6

6.5

6.8

6.3

6.2

6.3

6.4

6.8

7.0

6.3

4.0

3.4

3.5

3.6

3.7

3.8

3.9

4.2

4.4

3.7

7.9

7.8

7.9

8.0

8.1^b

8.2

7.9

11.4

-3.8

-3.1

-3.0

-2.8

-2.9

-2.8

-3.0

-3.4

-4.5

-5.1

-2.9

-28.3

-10.7

-10.5

-10.4

-10.2

-9.7

CtC(C₆H₄)NO₂(1a⁺)

CtC(C₆H₄)CN (1b⁺)

CtC(C₆H₄)CF₃(1c⁺)

CtC(C₆H₄)Br (1e⁺)

CtC(C₆H₄)F (1d⁺)

CtC(C₆H₅) (1f⁺)

-25.8

-29.0

-31.2

-40.2

-43.8

-41.7

-48.2

-55.3

-66.2

-67.4

-44.4

26.6

26.7

26.8

29.0

27.7

29.2

30.0

26.9

21.4

10.0^b

0.8

1.0

1.2

1.5

1.8

2.2

2.7

3.6^b

4.2^b,c

2.0

7.2^b,c

7.4^b,c

7.7^b,c

n.o.^d

n.o^d

n.o.^d

7.4^b,c

8.2^b,c

9.8^b

10.7^b

n.o.^d

-41.7 (H_p)^b,c

69.3 (p-Me)

15.3 (p-OMe)

-3.8 (NH)^b

78.9^b

-13.3 (m-Me)

-46.5 (H_p)

76.9 (p-Me)

47.4 (o-Me)

11.0 (H_m)

CtC(C₆H₄)CH₃(1g⁺)

CtC(C₆H₄)OCH₃(1h⁺)

CtC(C₆H₄)NH₂(1i⁺)

CtC(C₆H₄)NMe₂(1j⁺)

CtC-3,5-Xyl (3a⁺)

n.o^d

n.o.^d

-9.4

-10.3

CtC-2,4,6-Mes (3b⁺)

CtC(C₆H₅)Ph (4⁺)

33.7

30.9

7.6

6.3

6.8

6.7

6.3

2.7

1.6

4.0

3.7

8.1

8.0

n.o.^d

-3.4

-2.8

-10.5

-10.3

-44.8

n.o.^d

-0.4 (H_o)

-0.8 (H_p)

CtC-3-(C₆H₄F)

(m-1d⁺)

-34.1

-34.7

26.1

n.o.^d

6.9

6.2

1.4

3.6

7.9

7.4^a,b

-2.7

-10.5

-37.0 (p-H)

^aProposed assignment according to Chart 2 (CHDCl₂at 5.35 ppm). ^bTentative assignment. ^cPartly hidden behind other signals. ^dNot observed.

Scheme 2. Five-Membered Ring Interconversions of the

FeP₂(CH₂)₂Core

are also the least reactive in solution,²⁹were therefore used for

optimization of the measurement conditions.

For all Fe(III) compounds, the C₅Me₅protons (H₁₁) cor-

respond to the most intense peak at ca. -10 ppm. These protons

Figure 3. ¹H NMR spectrum of 1a[PF₆] in CD₂Cl₂at 25 °C with

proposed assignment according to Chart 2.

exhibit the largest shifts after the protons on the arylethynyl

ligand. Regarding the dppe ligand, two methylene signals are

observed at 11.4 and -3.8 ppm for 2[PF₆], which correspond

to the two protons endo (near) and exo (remote) from the

acetylide bond (Scheme 2). For 1a-j[PF₆], the high-field dppe-

methylene peak is slightly downfield shifted and observed near

-3 ppm as a broad singlet. However, the second methylene

signal is much more difficult to detect, since it is broader and

appears at lower fields. As a result, it is usually hidden beneath

more intense signals in the aromatic region. The existence of

this signal, which relaxes much faster than other signals in the

aromatic region, was ascertained by measurements of spin-

lattice relaxation time with 1a[PF₆] (T₁) 2.3 ms; Supporting

Information). Notably, in a frozen C_sconformation of the

metalladiphosphane five-membered ring made by the chelate

backbone coordinated to the iron center, each of the four

methylenic protons should be distinct. However, rapid inter-

conversions of the five-membered ring in solution result in an

effective C_2Vsymmetry for the “(η²-dppe)(η⁵-C₅Me₅)Fe” core

(Scheme 2), and the diagnostic AB pattern expected for the

inequivalent ³¹P nuclei in such a frozen ring is never observed

by ³¹P NMR, as can be stated with the diamagnetic Fe(II) parents

1a-j.³⁰

Chart 2. ¹H Nuclei Numbering Corresponding to the

Proposed Assignment

ppm), while the para methyl group of 3b[PF₆] is strongly shifted

downfield (79.9 ppm), as observed for 1g[PF₆] (69.3 ppm).

Remarkably, the signals of the ortho (H₁) and meta (H₂) ring

protons appear significantly influenced by the para substituents

in 1a-j[PF₆]. Thus, as X becomes more electron-releasing, these

peaks are more shifted and broadened. Given the increasing

reactivity of compounds with electron-releasing groups, a clean

spectrum of the amino-substituted complexes 1i[PF₆] and

1j[PF₆] proved challenging to obtain. In contrast, the most

resolved spectra were obtained with compounds presenting

electron-withdrawing substituents such as the nitro (1a[PF₆])

or the cyano (1b[PF₆]) complexes. The latter compounds, which

Concerning the aromatic protons of the dppe ligand, two sets

of signals in a roughly 1/2/2 ratio and with increasing linewidths

are observed. These can be readily assigned to para, meta, and

(30) Internal motions can be quite rapid, since variable-temperature NMR

experiments on 1a, 1g, and 3b revealed that no specific conformation

could be frozen in solution with these compounds at temperatures down to

190 K.

Spin Delocalization in Electron-Rich Iron(III) σ-Acetylides

Organometallics, Vol. 26, No. 4, 2007 879

Chart 3

in the 120-160 ppm region. Their exact assignment was subse-

quently obtained from the ¹H assignment with the help of

1

HMQC correlations (Table 2). As already observed in the H

NMR spectra, the peaks of the ortho carbon nuclei are signifi-

cantly broader than those of the meta and para carbon atoms.

In the ¹³C NMR spectra, these signals are also more shifted

than those of the meta and para carbon atoms. Notably, the peak

corresponding to the methylene carbon atoms of the dppe,

observed near -190 ppm in all compounds, is not detected with

2[PF₆].³¹

Regarding the arylethynyl carbon nuclei, the signals of the

acetylide R-carbon atoms (C₁) escaped detection for the Fe(III)

acetylide complexes presently investigated (Figure 4), in contrast

to the peaks of the â-carbon atoms (C₂), which were most often

detected. These signals constitute the most downfield-shifted

peaks of the various ¹³C NMR spectra (Figure 5). In this

connection, the most upfield-shifted peaks among the signals

detected for the arylethynyl ligands in 1a-h[PF₆] correspond

to the nearby carbon atom (C₃), in para-position to the

X-substituent. These very diagnostic peaks become broader for

the Fe(III) compounds with more electron-releasing substituents

and eventually become undetectable in 1i[PF₆] and 1j[PF₆]. As

already observed on the ¹H NMR spectra, the ¹³C NMR

chemical shifts of the arylethynyl ligand are quite substituent-

sensitive for 1a-i[PF₆] and are overall larger than those

undergone by the carbon nuclei on other ligands.

ortho protons, respectively (Supporting Information), the latter

being possibly in part broadened by unresolved coupling to

phosphorus. Similar to the methylenic dppe signals H₉and H₁₀,

these two sets correspond to phenyl rings “close” (endo) and

“remote” (exo) from the acetylide bond, respectively (Chart 2).

Notably, the peaks attributed to the phenyl protons H₃, H₄, and

H₅and to the methylene proton H₉correspond to the most

broadened, but also to the least shifted signals among those of

the dppe aromatic protons.

We next recorded the ¹H NMR spectra of m-1d[PF₆] and of

the known 4[PF₆]^10acomplexes (Chart 3). For the former

complex, some uncertainty remains regarding the assignment

of the ortho and para protons on the fluorinated aryl ring, both

signals being shifted to high fields (Table 1). We propose that

the two signals at high field near -34 ppm correspond to the

two nonequivalent ortho protons, while the more shifted signal

near -37 ppm would correspond to the para proton. The

nonequivalence of the two ortho protons must be weak, because

the corresponding carbon atoms are also not differentiated in

the ¹³C NMR (see below). For 4[PF₆], all the new peaks can be

attributed on the basis of their integrations and shifts (Table 1).

Assignment of the ¹³C NMR Spectra. The ¹³C NMR spectra

of all the Fe(III) compounds were next recorded. As for the ¹H

NMR spectra, we started by monitoring the shifts of mixtures

of 1a and 1a[PF₆] in order to assign the various signals detected

(Figure 4). Owing to the rapid self-exchange reaction, an

averaged set of peaks was always obtained for the various

mixtures tested. All observed signals could be identified. Thus,

the peaks corresponding to the two ipso carbon atoms of the

dppe aromatic rings (C₇and C₁₁) and to the R-carbon of the

alkynyl ligand (C₁) are lacking, while the peaks of all other

carbon atoms were detected in the 900/-200 ppm range. These

signals have already disappeared when only 5% of Fe(III) has

been added to the solution of corresponding Fe(II) complex.

More generally, with 1a-j[PF₆] we were unable to detect any

peaks outside this spectral range, even up to 4000 ppm and down

to -1000 ppm. Given the similarities between spectra, assign-

ments similar to those of 1a[PF₆] were taken for other

compounds (Table 2). The typical ¹³C NMR spectrum of 1a[PF₆]

is shown in Figure 5, along with the corresponding assignment

of the carbon nuclei (Chart 4).

Again, a quite specific signature is observed for the “(η²-

dppe)(η⁵-C₅Me₅)Fe” core, which is exemplified by the ¹³C NMR

spectrum of 2[PF₆]. Depending on the complex considered, the

singlet corresponding to the methyl groups of the C₅Me₅ligand

(C₁₇) is shifted downfield (2[PF₆]) or upfield (other compounds)

in comparison to its position in the corresponding Fe(II) com-

plex, while the peak of the inner carbon atoms of the cyclo-

pentadienyl ligand (C₁₆) is always shifted to low field. The latter

is detected at 223.3 ppm for 2[PF₆] and lies around 175 ppm

for 1a-j[PF₆] (Table 2). Regarding the dppe ligand, the signals

of the quaternary ipso carbons of the two nonequivalents phenyl

rings of the dppe ligand were not detected, while those

corresponding to the various primary carbon nuclei are clustered

The ¹³C NMR spectra of the m-1d[PF₆] and 4[PF₆] complexes

were next assigned on the basis of these results and of the

relative intensity of the detected signals (Table 2). For the former

compound, only one additional peak relative to 1d[PF₆] was

detected. The unique singlet appearing at high field certainly

corresponds to the overlapping signals of the two chemically

nonequivalent ortho C-H carbon nuclei of the fluorinated aryl

ring, while the C-H meta carbon nucleus is observed as

expected, near -33 ppm. The C-F meta carbon atom is observed

near 20 ppm, as a poorly resolved multiplet with a doublet-like

shape (¹J_CFof ca. 240 Hz in m-1d).²⁸For 4[PF₆], three additional

signals were detected relative to 1f[PF₆]. Two of these, near

200 ppm, correspond to the ortho and para C-H carbon nuclei

of the phenyl substituent. As expected from their more remote

position from the metal center, they are less shifted than the

peaks corresponding to C₄and C₆(Chart 4). The remaining new

peak near -12 ppm is tentatively attributed to the ipso carbon

atom. The missing peak that corresponds to the two meta C-H

carbons is possibly hidden behind the ortho dppe signal near

96 ppm (C₁₂).

³¹P NMR Spectra. So far, ³¹P NMR signals have never been

detected between 500 and -500 ppm for all the Fe(III)

complexes of this kind ever investigated.^6aTo make sure about

the possibility to detect this signal, we have decided to monitor

the shift of the phosphorus atom using various 1a/1a[PF₆]

mixtures. Thus, we started from pure 1a (δ_P) 100.6 ppm) and

progressively increased the amount of 1a[PF₆] in the solution

(Figure 6). The sharp phosphorus signal of 1a near 100 ppm

readily broadened and was strongly shifted toward high fields

already when traces of 1a[PF₆] complex were admixed. Above

10%, the signal rapidly flattened and eventually disappeared in

the baseline, possibly because of the increasing contribution of

line-broadening effects due to the electron self-exchange process

(vide infra). By extrapolation, we could infer that the ³¹P NMR

(31) For 2[PF₆] qualitatively similar spectra featuring the same number

of peaks were observed in DMSO-d₆when the temperature of the sample

was varied from 25 to 50 °C (above this temperature, 2[PF₆] decomposes),

rendering any coincidental overlap of the missing ipso carbon and methylene

signals unlikely.

880 Organometallics, Vol. 26, No. 4, 2007

Paul et al.

Figure 4. Observed ¹³C NMR shifts for 1a/1a[PF₆] mixtures in CD₂Cl₂at 25 °C (assignment according to Chart 4). Aromatic carbon

nuclei of the dppe ligand are shown on the diagram at the right (b). For all fits R²> 0.99.

a

Table 2. ¹³C NMR Shifts (δ (0.1 ppm) recorded for Selected [(η²-dppe)(η⁵-C₅Me₅)FeR][PF₆] Complexes at 20 °C in CD₂Cl₂

compd

CtC-4-(C₆H₄)-

C₃C₄C₅

dppe

C₁₂

159.0 134.5 131.2 98.9 123.9 121.0

C₅Me₅

C₁₆C₁₇

n.o.^b223.3 22.2

3.6

X

R

C₂

C₆

C₈

C₉

C₁₀

C₁₃

C₁₄

C₁₅

C_X

Cl (2⁺)

CtC(C₆H₄)NO₂

817.5 -134.0 311.2

789.5 -130.7 299.2

791.3 -134.7 292.3

40.0 336.6 155.3 137.5 131.2 95.8 124.0 119.4 -188.7 181.7

32.0 307.2 154.5 136.9 131.0 96.3 123.8 119.8 -184.6 179.4

12.4 332.9 154.8 136.6 130.6 95.8 123.6 119.8 -186.1 177.9

(1a⁺)

CtC(C₆H₄)CN

(1b⁺)

2.7 -16.0

1.6

30.4^c

CtC(C₆H₄)CF₃

(1c⁺)

CtC(C₆H₄)Br (1e⁺) 762.7 -143.0 276.3

CtC(C₆H₄)F (1d⁺) 744.5 -146.9 267.8

-45.9 455.4 154.4 136.3 132.1 95.6 123.1 120.0 -188.3 174.5 -0.5

-84.6 467.9 154.5 136.2 129.5 95.5 122.7 120.1 -189.1 173.2 -2.1

-50.8 396.2 155.0 135.7 130.8 95.2 122.5 119.7 -189.1 173.0 -2.2

CtC(C₆H₅) (1f⁺)

CtC(C₆H₄)CH₃

(1g⁺)

774.4 -147.0 267.0

748.8 -152.4 255.2 -101.0 475.7 155.5 136.1 129.2 95.9 122.5 120.4 -189.5 170.7 -3.2 -112.5^c

CtC(C₆H₄)OCH₃

707.8 -158.7 239.8 -151.9 497.3 155.4 136.2 128.4 96.5 122.1 120.6 -189.3 167.8 -4.7

n.o.^b,d

(1h⁺)

CtC(C₆H₄)NH₂

(1i⁺)

n.o.^b

n.o.^b206.4

n.o.^bn.o.^b

n.o.^b

n.o.^b157.7 137.2 129.2 96.8 121.6 121.6 -187.9 161.0 -7.3

CtC(C₆H₄)NMe₂

n.o.^b158.9 137.9 129.1 101.4 121.4 122.4 -189.9 158.9 -9.0 -81.7^c

(1j⁺)

CtC-3,5-Xyl (3a⁺) 767.3 -150.4 258.4

-66.2 429.7 155.6 135.8 129.3 95.6 122.4 119.8 -189.7 171.2 -3.2

776.4 -160.1 239.0 -127.8 552.3 158.2 136.6 129.6 96.3 121.6 121.0 -198.7 163.5 -3.5 -82.7

61.9

CtC-2,4,6-Mes

(3b⁺)

-160.0^c

CtC(C₆H₅)Ph (4⁺) 760.0 -149.1 263.8

-88.1 461.3 155.1 136.0 129.3 95.8 122.6 120.1 -188.8 172.3 -2.4

-32.9 345.8 154.2 135.8 129.8 95.3 122.8 119.5 -187.1 175.0 -0.2

231.0^e

174.1^f

n.o.^g

-12.2^h

18.0^j

CtC-3-(C₆H₄F)

(m-1d⁺)

758.6 -139.6 281.1ⁱ

^aProposed attribution according to Chart 4 (CD₂Cl₂at 53.8 ppm). ^bNot observed. ^cTentative assignment (broad and weak signal). ^dPossibly hidden

beneath another signal. ^eOrtho carbon atoms. ^fPara carbon atom. ^gMeta carbon atoms. ^hIpso carbon atom. ⁱInequivalent ortho carbon atoms apparently

not differentiated by meta-substitution. ^jFunctional meta carbon (C-F).

signal for neat 1a⁺should lie around -3000 ppm. Actually,

this signal was detected as a broad singlet (ν_1/2of ca. 13 000

Hz) with a pure sample of 1a[PF₆], due to the absence of self-

exchange. However, due to the very bad baseline obtained for

such a large spectral window with our probe, phasing of the

spectra proved problematic and a large uncertainty certainly

exists for the shift of this signal. Related ³¹P NMR signals were

also detected around 3000 ( 300 ppm for 1f[PF₆], 1h[PF₆],

1i[PF₆], and 2[PF₆]. These shifts should however be considered

with caution in the absence of extrapolated estimates from

corresponding Fe(III)/Fe(II) correlations.

in the usual ¹⁹F NMR range (Table 3). Thus, in addition to the

expected doublet of the PF₆counterion near -74 ppm (¹J_PF

-

of ca. 710 Hz), a broad singlet (ν_1/2) 200 Hz) was observed at

-15.1 ppm for 1c[PF₆], i.e., downfield to the signal previously

reported for 1c at -61.5 ppm in CDCl₃.¹⁴For 1d[PF₆] a very

broad singlet (ν_1/2) 2500 Hz) was observed at 13.3 ppm in

CD₂Cl₂, i.e., also downfield to the ¹⁹F NMR peak of the Fe(II)

parent (-119.8 ppm), while for m-1d[PF₆], a comparably

narrower (ν_1/2) 160 Hz) singlet was detected at -137.4 ppm,

this time upfield to the signal of m-1d (-115.8 ppm).²⁸

Derivation of Selected Hyperfine Coupling Constants. As

shown in eq 1, the observed paramagnetic shift of a given

nucleus is the sum of a diamagnetic contribution (δ_dia), which

¹⁹F NMR Spectra. For the complexes 1c[PF₆], 1d[PF₆], and

m-1d[PF₆] in CD₂Cl₂, broad ¹⁹F NMR signals could be detected

Spin Delocalization in Electron-Rich Iron(III) σ-Acetylides

Organometallics, Vol. 26, No. 4, 2007 881

Figure 6. ³¹P NMR spectra of 1a/1a[PF₆] mixtures at 25 °C in

CD₂Cl₂for 0% (a), 5% (b), 12% (c), and 100% (d) of Fe(III)

complex (intensities are arbitrary). The signal of the PF₆^-counterion

can also be observed near -140 ppm. The observed line-broadening

of the signal upon increasing concentrations of Fe(II) complex

matches with that expected for the self-exchange reaction under

fast exchange conditions.

Figure 5. ¹³C NMR spectrum of 1a[PF₆] in CD₂Cl₂at 25 °C with

proposed assignment according to Chart 4. The two parts correspond

to two spectra with different offsets (intensities are arbitrary).

Chart 4. ¹³C Nuclei Numbering Corresponding to the

Proposed Assignment

Table 3. ¹⁹F NMR Shifts, Half-widths, Isotropic Shifts,

Estimated Isotropic Hyperfine Coupling Constants (G), and

Computed Pseudocontact Shifts (ppm) for Selected

[(η²-dppe)(η⁵-C₅Me₅)FeCtC-Ar_F][PF₆] Complexes at 20 °C

in CD₂Cl₂at 500 MHz

δ^a

(ppm)

ν_1/2

(Hz)

δ_iso

a_F

(δ_pc)^{M a,c}

(ppm)

a

b

compd

Ar_F

(ppm)

(G)

4-C₄H₄(CF₃)

4-C₄H₄F

3-C₄H₄F

-15.1

13.5

-137.5

200^d

2500^e

160^d

46.4

133.5

-20.5

0.49

1.44

-0.23

1.0

1.3

b

c

d

^a(0.5 ppm. Estimates obtained from eq 4. Obtained from eq 3. (10

e

Hz. (100 Hz.

result solely from the metal-centered term (δ_pc)^M. This contribu-

tion can be computed from the geometric parameters of the

complex, provided the magnetic anisotropy or the diagonal

values of the g-tensor are known. Thus, for 1a-j[PF₆], a

rhombic anisotropy should have been considered based on ESR

data.¹⁵However, considering (i) the very fluxional nature of

these complexes in solution (see Scheme 2 for instance), (ii)

the very facile rotation of the arylacetylide ligand around its

axis (see later), and (iii) the uncertainties regarding the precise

orientation of the g-tensor along with the overall weak rhom-

bicity of the g-tensor (g₃> g₂≈ g_e≈ g₁),³²we have chosen to

simplify the derivation of (δ_pc)^Mby considering an axial

anisotropy instead. Thus, we have used eq 3¹⁷with g_|) g₃

directed along the acetylide axis and g ) (g₂+ g₁)/2 to

compute the pseudocontact shifts for protons of the arylacetylide

ligand in which we were primarily interested (Table 4).³³As

can be seen from the values found, the metal-centered pseudo-

contact contribution to the shifts of these protons is quite weak

in 1a-j[PF₆] or 3a,b[PF₆], since the estimates obtained fol-

corresponds to the shift that the compound would present

without any unpaired electron, and the isotropic contribution

(δ_iso), which precisely represents the additional contribution to

the shift due to the presence of the unpaired electron(s) in the

compound.¹⁷For the Fe(III) complexes under investigation, the

former contribution (δ_dia) can be approximated by the shift of

the nucleus considered in the diamagnetic Fe(II) parent.

δ_obs) δ_dia+ δ_iso

(1)

The remaining isotropic shift (δ_iso) can also be split into two

contributions, called contact (δ_c) and pseudocontact (δ_pc) shifts

(eq 2a).¹⁷The pseudocontact (or dipolar) shift (δ_pc) results from

the through-space dipolar interaction between the nuclear spin

and the unpaired electron, while the contact contribution (δ_c)

results from Fermi coupling (through-bond interaction) between

the nuclear spin and the unpaired electron. This shift is directly

proportional to the local spin density in s-type atomic orbitals

(AOs) and allows derivation of the corresponding hyperfine

coupling constant, as defined for organic radicals. However, in

order to access the desired hyperfine coupling constant from

the isotropic shift, the pseudocontact contribution originating

from metal center (δ_pc)^Mneeds to be reliably estimated first.

(32) Analysis of the frontier molecular orbitals (MOs) in a previous DFT

study conducted on computationally simpler model compounds suggested

that the direction of the strongest diagonal component (g_zz) g₃) was roughly

along the acetylide axis in 1a-j[PF₆], but slight deviations were predicted

to take place, especially with strongly electron-withdrawing substituents.¹⁵

(33) In this expression, r_Mis the distance from the metal center to the

nuclei, θ is the corresponding angle with the acetylide axis, and the symbols

have their usual meaning.³⁴The geometric parameters (r_Mand θ) were

obtained from the corresponding X-ray structures after averaging the data

for equivalent nuclei in solution. The δ_pcvalues computed for several

complexes are given in Table 4. Notably, eq 3 is reliable for r_Mvalues

larger than 4 Å (less than 10% accurate if not).^35,40Thus, δ_pcvalues derived

by this equation should be considered with care for nuclei closer to the

metal center or for nuclei on very fluxional ligands (due to the inherent

imprecision on the geometric factors). This is, however, not the case for

the detected nuclei of the arylacetylide ligand.

δ_iso) δ_c+ δ_pc

(2a)

(2b)

δ_pc) (δ_pc)^M+ (δ_pc)^L

As shown in eq 2b, the pseudocontact shift can itself be

decomposed into two contributions. In complexes where the

unpaired electron mostly resides on the metal center, the ligand-

centered term (δ_pc)^Lis usually neglected for hydrogen nuclei.

Most often, the pseudocontact shift is therefore considered to

882 Organometallics, Vol. 26, No. 4, 2007

Paul et al.

Table 4. Experimental Paramagnetic Shifts (δ (ppm) ( 0.1),

Calculated Metal-Centered Pseudocontact Shifts (δ_pc), and

Resultant Contact Shifts (δ_c) for Arylacetylide Protons of

Selected [(η²-dppe)(η⁵-C₅Me₅)FeCtC-Ar][PF₆] Complexes

at 20 °C in CD₂Cl₂(5.35 ppm)

Table 5. Isotropic Hyperfine Coupling Constants (in G) with

Selected Protons for [(η²-dppe)(η⁵-C₅Me₅)FeR][PF₆]

Complexes

hyperfine constants

compd

Ar

X

R ) -CtC-4-(C₆H₄)X

a_H1

a_H2

a_HX

compd

Ar

CtC(C₆H₄)NO₂(1a⁺)

CtC(C₆H₄)CN (1b⁺)

CtC(C₆H₄)CF₃(1c⁺)

CtC(C₆H₄)Br (1e⁺)

CtC(C₆H₄)F (1d⁺)

-0.42

-0.45

-0.48

-0.59

-0.63

-0.61

-0.69

-0.78

-0.91

-0.93

-0.64

0.50^d

0.20

0.21

0.23

0.22

0.24

0.25

0.23

0.17

0.04

-0.19 ^c

0.29

0.20

H₁

H₂

H_X

(C₆H₄)NO₂

δ_i

-32.6

3.7

-36.3

-55.0

3.5

-58.5

-73.4

2.4

-75.8

-51.4

3.4

18.7

1.5

17.2

23.1

1.5

21.6

15.0

1.0

a

(1a⁺)

δ_pc

δ_c

δ_i

CtC(C₆H₅) (1f⁺)

-0.61^a

0.78

0.13

(C₆H₄)CH₃

67.1

0.8

66.3

-6.0

0.6

CtC(C₆H₄)CH₃(1g⁺)

CtC(C₆H₄)OCH₃(1h⁺)

CtC(C₆H₄)NH₂(1i⁺)

CtC(C₆H₄)NMe₂(1j⁺)

CtC-3,5-Xyl (3a⁺)

CtC-2,4,6-Mes (3b⁺)

CtC-3-(C₆H₄F) (m-1a⁺)

(1g⁺)

δ_pc

δ_c

δ_i

δ_pc

δ_c

δ_i

δ_pc

δ_c

δ_i

δ_pc

δ_c

-0.08

(C₆H₄)NH₂

0.93^b

(1i⁺)

-0.64^a

0.87^e

145.0

-15.5^b

0.6^b

-6.6

-53.5^c

1.1^c

3,5-Xyl

-0.53 ^f

(3a⁺)

-0.52

-0.53

-0.65

-54.8

45.3^d

2.5^d

-16.1^b

26.4

1.4

-54.6^c

74.7^e

0.8^e

CtC-2-(C₆H₄)-1,1′-(C₆H₂) (4⁺)

0.26

-0.10^g

0.03^h

2,4,6-Mes

(3b⁺)

42.8^d

25.0

73.9^e

b

c

^aProton in para position. Methyl groups on nitrogen. Methyl group in

^aComputed from eq 3 and crystallographic data. ^bMethyl group in meta

position. ^cProton in para position. ^dMethyl group in ortho position. ^eMethyl

group in para position.

d

e

meta position. Methyl group in ortho position. Methyl group in para

position. Proton in para position. Protons in ortho and para positions on

the second ring. Proton in meta position on the second ring.

f

g

h

lowing our approximation remain below 7% of the isotropic

shifts, except for the meta protons of 1a[PF₆], where they

amount to 12% of the isotropic shift. In regard to the relative

weakness of this contribution, eq 3 certainly constitutes a

convenient and straightforward means to get a fair estimate of

the pseudocontact shifts (δ_pc) for H₁and H₂in these compounds

and provides thereby an access to the corresponding contact

contributions (δ_c) via eq 2a.

Table 6. π-Charge (e) on the Primary Carbon Atoms of the

Arylacetylide Ligand as Deduced from McConnell and

Karplus & Fraenkel Equations for

[(η²-dppe)(η⁵-C₅Me₅)FeR][PF₆] Complexes³⁷

CtC-4-C₆H₄X

compd

R

a

(F^π)_C4

(F^π)_C5

(F^π)_C6

(F^π)_X

CtC(C₆H₄)NO₂

CtC(C₆H₄)CN

CtC(C₆H₄)CF₃

CtC(C₆H₄)Br

CtC(C₆H₄)F

0.019

0.021

0.022

0.027

0.029

0.028

0.031

0.035

0.040

0.029

-0.009

-0.010

-0.011

-0.010

-0.011

-0.012

-0.010

-0.007

-0.002

-0.008^c,d

-0.013

-0.009

(δ_pc)^M) (µ₀/4π)(µ_B/9kT)[S(S + 1)](g_|²- g )(3 cos²θ -

2

0.014^b

3

1)/r_M(3)

CtC(C₆H₅)

0.027

CC(C₆H₄)CH₃

CC(C₆H₄)OCH₃

CtC(C₆H₄)NH₂

CtC(C₆H₄)NMe₂

CC-3,5-Xyl

CtC-2,4,6-Mes

CtC-3-(C₆H₄F)

0.031^c

From the contact shifts, the hyperfine coupling constants

between the unpaired electron and the protons considered were

obtained using eq 4 (N ) H).¹⁷In this expression, g represents

the isotropic g value, S is its spin quantum number, T is the

temperature in Kelvin, while other symbols have their usual

meaning.³⁴The isotropic g values needed in eq 4 were obtained

by averaging the three diagonal g values experimentally

determined at low temperature by ESR for the various Fe(III)

complexes.¹⁵The values found for hydrogen and fluorine atoms

of the arylacetylide ligands in 1a-j[PF₆], m-1d[PF₆], 3a,b[PF₆],

and 4[PF₆] are given in Tables 3 and 5. Given the approxima-

tions made, these figures are evidently only estimates of the

exact hyperfine coupling constants with these nuclei. Note

however that due to the poor resolution achieved during ESR

measurements on frozen glass samples, such data could not be

formerly obtained.

-0.003^c

0.029

0.034^c,f

0.023

0.020^c,e

0.024

0.029

CtC-4-(C₆H₄)Ph

-0.012

0.004^g

-0.001^h

0.004^g

b

^aDetermined from δ_Cand eq 5. A value of 107.6 MHz has been used

F

for (Q_C-CF3

)

with the ¹⁹F NMR isotropic shift corrected for the metal-

centered dipolar contribution.³⁹^cA value of 75 MHz has been used for

H

(Q_C-CH3

)

and for (Q_C-NH2)^H. ^dMethyl group in meta position. ^eMethyl

groups in ortho position. ^fMethyl group in para position. ^gProtons in ortho

and para positions on the second ring. ^hProton in meta position on the

second ring.

in the p_zatomic orbital (AO) of the carbon atom bearing the

hydrogen nucleus.^18,36The latter can be deduced using the so-

called McConnell equation (eq 5), where (Q_CH)^His a constant

amounting to ca. -66 MHz (Table 6).^36b,37For methyl groups,

a_N) δ_c× p3γ_NkT/(gµ_BS(S + 1))

(4)

Derivation of Spin Densities on Selected Nuclei. According

1

to McConnell, the contact contribution to H NMR shifts of

aromatic radicals is directly proportional to the local spin density

(36) (a) McConnell, H. M.; Chesnut, D. B. J. Chem. Phys. 1958, 27,

984-985. (b) McConnell, H. A. J. Chem. Phys. 1958, 28, 1188-1192.

(37) Karplus, M.; Fraenkel, G. K. J. Chem. Phys. 1961, 35, 1312-

1323.

(38) Mispelter, J.; Momenteau, M.; Lhoste, J.-M. In Biological Magnetic

Resonance; Berliner, L. J., Reuben, J., Eds.; Plenum Press: New York,

1993; Vol. 12, pp 299-355.

(39) Eaton, D. R.; Josey, A. D.; Sheppard, W. J. Am. Chem. Soc. 1963,

85, 2689-2694.

(34) µ₀is the vacuum permitivity, γ_Nare the magnetogyric ratio of the

nuclei N, µ_Bis the Bohr magneton, k is the Boltzman constant, p is the

reduced Planck constant, g_eis the free electron g value, and N is the

Avogadro number.

(35) (a) Golding, R. M.; Pascual, R. O.; Vrbancich, J. Mol. Phys. 1976,

31, 731-744. (b) Golding, R. M.; Pascual, R. O. Bull. Magn. Reson. 1983,

5, 126-128.

Spin Delocalization in Electron-Rich Iron(III) σ-Acetylides

Organometallics, Vol. 26, No. 4, 2007 883

π(ν_1/2) ) (T₂*)^-1

constants of 75 MHz have been used in 1g[PF₆] and 3a,b[PF₆].

(6)

(a_H)^π) (Q_CH)^H(F^π)_C

(5)

The T₂* values previously determined for the various protons

in 1a[PF₆] or 1b[PF₆] according to eq 6 are clearly shorter than

the longitudinal relaxation rates (T₁, Supporting Information).

Considering T₂) T₂* (i.e., neglecting any contribution to the

linewidths due to the magnetic field inhomogeneity), the longer

values found for T₁in comparison to T₂* indicate that we are

not in the so-called fast motion limit.¹⁷Consequently, the T₂

values determined for protons in 1a-j[PF₆] can be approximated

by eq 7a (see also Supporting Information). This expression is

a sum of different terms corresponding to distinct relaxation

mechanisms.⁴⁴The first term pertains to dipolar relaxation and

the second to contact relaxation.⁴⁵In this expression, the

correlation time (τ_c) of the dipolar relaxation is determined by

the rotational correlation time (τ_r) and by the electronic

correlation time (τ_e) according to eq 7d. Examination of the

field dependence of the half-widths of the various peaks in the

¹H NMR spectra for 1a[PF₆], 1g[PF₆], and 2[PF₆] over a 200,

300, and 500 MHz apparatus reveals a poor sensitivity to the

field strength (Supporting Information). This suggests that the

transverse relaxation rates are not dispersed owing to 1/(1 +

τ²ω_s²) terms (τ ) τ_cor τ_e) and, therefore, that these terms might

be neglected in the spectral density functions (eqs 7b and 7c).

While the carbon spin densities obtained by these means can

be considered with confidence for primary carbon atoms in

complexes with electron-withdrawing substituents, the spin

densities obtained for the carbons bearing electron-releasing

substituents are certainly less reliable because of the neglect of

the contribution of local dipolar shifts in our treatment (see

below). Also, given the known variability of the corresponding

Q factor, for which several values have been computed or

measured depending on the compounds, the charge derived for

the few quaternary carbons bearing a methyl substituent should

also be considered with great care.¹⁷

1

Temperature Dependence of the H NMR Signals. We

have next examined the temperature dependence of the ¹H NMR

shifts of 1a[PF₆], 1g[PF₆], and 2[PF₆] between 297 and 183 K.

For each peak, the temperature dependence of the isotropic shift

(δ_iso) provides information about the energetic degeneracy of

the ground state (GS) of the paramagnetic compound under

investigation and on its separation from the first excited states.¹⁷

As shown in eq 2a, the isotropic shift is the sum of the contact

(δ_c) and pseudocontact (δ_pc) shifts. For paramagnetic compounds

with a nondegenerate GS and high-lying excited states, the

isotropic shift should exhibit a 1/T dependence converging

toward zero for T reaching infinity (Curie behavior), since both

the contact (δ_c) and the pseudocontact (δ_pc) contributions are

expected to follow such a trend.^17,40As shown in Figure 7 for

1g[PF₆], linear plots were obtained against 1/T with overall very

good fits for all signals in the temperature range investigated

(Table 7). This linear dependence is clearly in line with the

Curie behavior expected for these compounds. Except for the

ortho protons (H₁) of the arylalkynyl ligand in 1a[PF₆] and for

the C₅Me₅protons (H₁₁) in 2[PF₆], the shifts of the protons in

the different complexes correspond quite satisfyingly to those

of the diamagnetic Fe(II) parents (δ_diain eq 1) when extrapolated

at infinite temperature, especially when considering the inher-

ently large error of such a procedure (Table 7). Thus, we believe

that the discrepancies observed for H₁in 1a[PF₆] and H₁₁in

2[PF₆] are not indicative of a GS degeneracy, nor of closely

lying excited states. We would rather tentatively propose that

they result from the stabilization of some conformations, solvent

adducts, or aggregates presenting less spin density on H₁or

H₁₁when the temperature is decreased.⁴²

2

(T_2M

)

^-1) (1/15)(µ₀/4π)²(γ_Hg²µ_BS(S + 1)/r_M⁶)f_dip(ω,τ_c) +

(1/3)S(S + 1)(a_H/p)²f_con(ω,τ_e) (7a)

f_dip(ω,τ_c) ) 7τ_c+ [13τ_c/(1 + τ_c²ω_s²)]

(7b)

(7c)

(7d)

f

_con(ω,τ_s) ) τ_e+ [τ_e/(1 + τ_e²ω_s²)]

(τ_c)^-1) (τ_e)^-1+ (τ_r)^-1

Also, it seems that no particular mechanism dominates the

relaxation process of all protons in these compounds.⁴⁶Most

probably, an interplay of the dipolar and contact relaxation

operates, as often observed with paramagnetic complexes.

However, because of the (r_M)^-6dependence of the metal-

centered dipolar relaxation, the contact relaxation can be

expected to become the dominant relaxation process for protons

situated at the periphery of the compounds, provided sufficient

spin density is delocalized to these nuclei. In line with this

supposition, a reasonably linear fit can be obtained between the

squared isotropic displacements and the half-widths of the ortho

protons (H₁) on the arylacetylide ligand in 1a-g[PF₆] (Figure

8). For 1i[PF₆] (X ) OMe) and complexes with more electron-

releasing substituents, the data start to deviate from this line

Electronic and Rotational Correlation Times Deduced

from the ¹H NMR Signals. The half-widths (ν_1/2) of the various

NMR signals were also measured. These data are inversely

proportional to the transverse relaxation rates (T_2M) (eq 6) and

convey therefore important information about the relaxation

processes operative in solution.

(44) (a) Ko¨hler, F. H. Z. Naturforsch. 1974, 29B, 708-712. (b) Banci

L. In Biological Magnetic Resonance; Berliner, L. J., Reuben, J., Eds.;

Plenum Press: New York, 1993; Vol. 12, pp 79-111.

(45) In eq 7a, many symbols were already defined before:³⁴r_Mis the

distance to the metal center in Å, ω_sis the spin angular frequency, f_dip

-

(ω,τ_c) and f_con(ω,τ_e) represent the spectral density functions where τ_cand

τ_erepresent the various correlation times considered, while a_His the

hyperfine coupling constant with the proton under investigation (in J). Note

also that Curie-type relaxation processes have been neglected in equation

7a, since they are often negligible for small complexes such as 1a-j[PF₆].

Accordingly with this hypothesis, any line-broadening computed for a Curie-

type dipolar relaxation process appears to be negligible when τ_rvalues are

used as maximum estimates of τ_c.¹⁷

(40) Regarding the expression of δ_pc(eq 3), the following conditions

are apparently verified: τ_r^-1< |g_|- g |µ_BHp^-1and τ_e, τ_r(see later).⁴¹

(41) Jesson, J. P. J. Chem. Phys. 1967, 47, 579-581.

(42) Indeed, a clear deviation of the Curie law is not obserVed for all

signals, as could be expected if temperature-independent and/or 1/Tⁿ-

dependent (n > 1) terms were present in the expression of the paramagnetic

shift due to GS degeneracy of closely lying excited states.^35aMoreover,

DFT computations on the various model complexes predict a nondegenerate

ground state (GS) in which the gap between the singly occupied molecular

(46) The lack of clear (field)²-dependence for all ν_1/2values excludes a

dominant Curie-type dipolar relaxation as supposed while deriving equation

7a. Then, the nonconstant ratio between ν_1/2values for similar protons in

two different complexes excludes a dominant dipolar relaxation process,

while the lack of proportionality between the ν_1/2values and the corre-

sponding squared contact shifts excludes a dominant contact mechanism.

orbital (SOMO) and the closest MO is larger than 2000 cm^{-1 15}

contradiction with low-lying excited states.⁴³

,

also in

(43) (a) Golding, R. M. Pure Appl. Chem. 1972, 32, 123-135. (b)

Kurland, R. J.; McGarvey, B. R. J. Magn. Reson. 1970, 2, 286-301.

884 Organometallics, Vol. 26, No. 4, 2007

Paul et al.

1

Figure 7. Temperature dependence of the H NMR shifts of 1g[PF₆] in CD₂Cl₂with proposed assignment according to Chart 2.

1

Table 7. Comparison between the H NMR Shifts (δ_dia) Recorded for 1a, 1g, and 2 at 20 °C and Those Extrapolated (δ_∞) from

Shifts vs Temperature Plots Obtained with 1a[PF₆], 1g[PF₆], and 2[PF₆] in CD₂Cl₂

CtC-4-(C₆H₄)

dppe

H₇

C₅Me₅

H₁₁

X

compd

H₁

H₂

H₄

7.4

6.4

0.97

7.0-7.5

6.6

0.81

H₅

7.4

6.8

0.99

7.0-7.5

7.0

0.99

H₈

7.5

6.9

0.99

7.0-7.5

7.1

0.99

H₁₀

H_X

a,b

1a

1a[PF₆]

δ_dia

6.8

7.9

7.3

2.1

5.5

0.99

2.0

2.6

0.99

2.0

3.0

1.5

0.2

0.99

1.4

-1.0

0.99

1.1

c

δ_∞

-9.2

0.99

10.6

0.99

R^{2 d}

0.99

7.0-7.5

7.0

0.99

7.1-7.5

7.2

a

1g

1g[PF₆]

δ_dia

6.8

6.9

2.2

4.7

0.99

c

δ_∞

4.4

7.7

R^{2 d}

0.99

a

2

δ_dia

7.1-7.5

6.9

0.97

7.1-7.5

7.4

0.99

7.1-7.5

7.4

0.99

c

2[PF₆]

δ_∞

-8.5

0.99

R^{2 d}

0.99

^{a 1}H NMR shift at T ) 295 K for the corresponding Fe(II) compound. ^bComplete attribution made by COSY and NOESY experiments on a 500 MHz

1

spectrometer. ^cExtrapolated H NMR shift at T ) ∞. ^dSquared regression coefficients of the linear fits.

and are more accurately fitted by a logarithmic law, the

compound 1g[PF₆] (X ) Me) constituting the limiting case for

the linear dependence. This “logarithmic” dependence observed

in Figure 8 most probably takes place because we neglected

the dipolar contribution to the shift induced by local electron

density in our treatment ((δ_pc)^Lin eq 2b). The latter becomes

certainly quite important for the electron-releasing substituents,

since quite sizable electronic densities are delocalized on the

functional aryl ring (see DFT calculations section). As a con-

sequence, the contact shift is derived in a more and more approx-

imate way. Moreover, any local dipolar contribution to the line

broadening was also neglected in eq 7a. These approximations

certainly explain the nonlinear plot of Figure 8.

We next sought to obtain an estimate of the dipolar correlation

time (τ_c) in eq 7a at ambient temperature with 1a[PF₆], 1g[PF₆],

and 2[PF₆]. An estimate of the rotational correlation time for

1a-j[PF₆] and 2[PF₆] can be readily obtained using the Stokes-

Einstein formula (eq 8), where η is the viscosity of the medium

(0.423 10^-3kg m^-1s^-1at 298 K) and a is the mean radius of

the molecule (8.46 and 10.36 Å, respectively). Values of τ_r)

4.8 × 10^-10s and τ_r) 2.6 × 10^-10s were respectively found

for 1a-j[PF₆] and 2[PF₆].

by contact relaxation. In line with a dominant dipolar relaxation

mechanism, the half-widths of these ¹H NMR signals vary only

slightly along the 1a-j[PF₆] series. The computation of the

dipolar line-broadening reveals that much larger half-widths

should be experimentally observed, evidencing that the actual

dipolar correlation times (τ_c) in 1a[PF₆], 1g[PF₆], and 2[PF₆]

must be lower than the rotational correlation times (τ_r) presently

considered. In these compounds, τ_cappears therefore signifi-

cantly influenced by τ_e(eq 7d). We found that dipolar correlation

times around 7 × 10^-11s would produce the observed half-

widths of H₅and H₈in 1a[PF₆], 1g[PF₆], and 2[PF₆]. From these

values, τ_evalues slightly below 8.5 × 10^-11s can be inferred,

but these are most probably high-lying estimates. Indeed, the

half-widths computed using the τ_evalues along with τ_rvalues

previously determined (eq 8) are again larger than those

experimentally observed for some other peaks, meaning that

the correlation time (τ_c) of the dipolar relaxation process must

be even shorter for these nuclei. The only way to reconcile this

observation with experiment is that either (i) our supposition

that the relaxation of H₅and H₈is mostly dipolar in origin was

wrong or (ii) that a different “τ_e” operates for different protons.

This can only be possible if the latter is in turn dependent on

an “internal” correlation time (eq 9).^44bInternal correlation times

(τ_i), which can be as low as 10^-12s,^30,47depend on intramo-

lecular motions and can be different for various protons located

on different fluxional groups. Whatever the exact reason, the

τ_evalues determined above constitute major values of τ_sin 1a-

j[PF₆] according to this reasoning. More sensible estimates can

τ_r) 4πηa³/3kT

(8)

Then, still using these τ_rvalues as maximum estimates of τ_c

along with r_Mdistances derived from X-ray structures, we have

computed the theoretical line-broadening induced by dipolar

relaxation for the two protons H₅and H₈of the dppe ligand in

1a[PF₆], 1g[PF₆], and 2[PF₆]. The peaks of these protons present

the smallest shifts and should therefore be only poorly broadened

(47) Jesson, J. P. J. Chem. Phys. 1967, 47, 582-591.

Spin Delocalization in Electron-Rich Iron(III) σ-Acetylides

Organometallics, Vol. 26, No. 4, 2007 885

be obtained from the half-widths of H₁(see Discussion section).

(τ_e)^-1) (τ_s)^-1+ (τ_i)^-1

(9)

Note that because of the much more important contribution

of local dipolar effects on relaxation of other nuclei, similar

information cannot be extracted so easily from the half-widths

of the ¹³C NMR and ¹⁹F NMR signals. With these nuclei, much

more complex expressions than eq 7a have to be considered in

the general case.^38,48

Self-Exchange Rates from Line-Broadening. As mentioned

above, we were always under fast exchange conditions when

making assignments from H, ¹³C, and ³¹P NMR spectra from

Fe(II)/Fe(III) mixtures. We therefore have sought to determine

the self-exchange rates from line-broadening studies using Fe-

1

Figure 8. Correlation between ν_1/2(Hz) values and squared

1

isotropic shifts (ppm²) for H NMR signals of H₁protons in 1a-

j[PF₆] complexes at 25 °C in CD₂Cl₂.

(II)/Fe(III) mixtures (Table 8). We used the classical formula

Table 8. Self-Exchange Rates (10^-6M^-1s^-1) as Determined

-1 51

given in eq 10a, which applies when ck_e. (∆δ)²/ck_e. T_2M

.

1

from Line-Broadening Studies on the H NMR Spectra for

In this expression, c is the total concentration of reactants, ∆δ

is the difference in chemical shift between the two nonexchang-

ing peaks expressed in Hz, and k_eis the self-exchange rate to

be determined. To avoid any errors due to the temperature

dependence of the Fe(III) proton shifts, ∆δ values were

estimated from spectra recorded at similar temperatures with

pure samples of the Fe(II) and Fe(III) redox congeners. The

Selected Compounds among 1a-g[PF₆] and 2[PF₆] in CD₂Cl₂

a

k_e(10^-6M^-1s^-1

)

∆H^q

compds

193 K

253 K

293 K

(kJ mol^-1/cm^-1

)

1a/1a[PF₆]^b

1b/1b[PF₆]^c

1c/1c[PF₆]^c

1g/1g[PF₆]^b

1h/1h[PF₆]^c

2/2[PF₆]^b

24

28

10

12

11

157

76

61

81

85

258

134

103

132

133

41.8

11.4/953

9.6/803

11.2/936

11.6/970

12.1/1011

8.6/718

self-exchange rates were determined from ca. (3-4) × 10^-2

M

solutions of the compounds in CD₂Cl₂, at several temperatures

in the 300-190 K range.

23.9

^aValues ( 15%. ^bDetermined at 300 MHz. ^cDetermined at 500 MHz.

W_red,ox) f_redf_ox4π(∆δ)²/ck_e

(10a)

(10b)

f_red) (δ_ox- δ_obs)/(δ_ox- δ_red) and f_ox) 1 - f_red

ments were repeated at various temperatures, showing that the

rates are slower at low temperatures. Activation energies around

11-12 kJ could be extracted from Eyring plots in each case

(Table 8). Rates for 2[PF₆] could not be determined below 230

K due to the excessive broadening of the C₅Me₅signal,

rendering the data extracted from Eyring plots less accurate for

that particular compound.

The rates were found to be in the range ca. 10.3 × 10⁷to

25.8 × 10⁷M^-1s^-1at ambient temperatures and are slightly

substituent dependent, being apparently larger for electron-

withdrawing substituents (Table 8). Notably, while eq 10a

applies in principle only to noncoupled protons,⁵²very similar

results were presently found for most compounds either from

1

Correlations of H and ¹³C NMR Shifts with Electronic

Substituent Parameters (ESPs). During previous studies on

1a-j[PF₆], we have shown that many characteristic properties

of these compounds, such as ESR data,¹⁵can be linearly

correlated using the σ⁺electronic substituent parameters

(ESPs).^53,54Again, if we except the data gathered for 1j[PF₆],

which most often is remote from the fit, rather similar linear

dependences appear with isotropic (Figure 9a) or contact shifts.

However, fits obtained here are overall poorer than those

obtained with ESR data. This is not surprising considering the

quite horizontal slopes of these linear correlations with respect

to the experimental uncertainty on the shifts. In many cases,

good correlations are also obtained using directly the uncorrected

paramagnetic shifts measured against tetramethylsilane instead

of the isotropic shift, as shown for the ¹³C NMR shifts of the

arylacetylide ligand in Figure 9b. As previously stated with ESR

data, slightly poorer fits (not shown here) were obtained when

the regular Hammett set was used instead of σ⁺, suggesting

the importance of mesomeric effects in these linear free-energy

relationships (LFERs).

1

the H NMR peaks of the coupled H₁and H₂protons or from

those of the C₅Me₅peak. This is not surprising considering the

rather large experimental uncertainty associated with these

measurements due to inavoidable field inhomogeneities ((10%).

We have also checked that these rates were independent of the

counterion concentration.⁵²Thus, similar rates were found for

2/2[PF₆] mixtures when a fixed concentration (0.08 M) of [Nⁿ-

Bu₄][PF₆] was present in the reaction medium. The measure-

(48) We have nevertheless checked that neither the dipolar nor the contact

relaxation dominated the shifts of the ¹³C NMR signals in 1a-j[PF₆]. In

the case of dominant metal-centered relaxation, we should have observed

a constant linewidth for each signal of these carbon nuclei along the 1a-

j[PF₆] series, given the relative constancy of the geometric parameters.

Alternatively, in the case of a dominant contact relaxation, a constant ratio

depending on their respective contact shifts (ν_1/2(¹³C_y)/ ν_1/2(¹H_x) ≈ (δ_C-

(¹³C_y)/δ_C(¹H_x))²×(γ_C/γ_H)²) should be found for a given couple of peaks.⁴⁹

No such ratios were found when the ¹³C NMR contact shifts were

approximated by the corresponding isotropic shifts. Rather, the analysis of

half-widths of the ¹³C NMR signal of the arylacetylide ligand reveals the

existence of rough correlations with the corresponding ¹³C NMR isotropic

shifts or with these quantities squared, in line with what is to be expected

when the transverse relaxation rates are determined by either local dipolar

effects or contact effects, or both.⁵⁰

Theoretical (DFT) Spin Densities for Selected Fe(III)

Model Complexes. Density functional theory (DFT) computa-

tions on Fe(III) were previously made on the model complexes

1a-H⁺(X ) NO₂), 1b-H⁺(X ) CN), 1d-H⁺(X ) Br), 1f-H⁺

(49) Doddrell, D. M.; Gregson, A. K. Chem. Phys. Lett. 1974, 29, 512-

515.

(50) (a) Doddrell, D. M.; Pegg, D. T.; Bendall, R.; Gottlieb, H. P. W.

Chem. Phys. Lett. 1976, 39, 65-68. (b) Doddrell, D. M.; Healy, P. C.;

Bendall, R. J. Magn. Reson. 1978, 29, 163-166.

(51) Simonneaux, G.; Bondon, A. Chem. ReV. 2005, 105, 2627-2646.

(52) Chan, M.-S.; Wahl, A. C. J. Phys. Chem. 1978, 82, 2542-2549.

(53) March, J. AdVanced Organic Chemistry. Reactions, Mechanisms and

Structures, 4th ed.; J. Wiley & Sons: New York, 1992.

(54) Hansch, C.; Leo, A.; Taft, R. W. Chem. ReV. 1991, 91, 165-195.

886 Organometallics, Vol. 26, No. 4, 2007

Paul et al.

1

Figure 9. Plot of the H NMR isotropic shifts (a) and the observed ¹³C NMR shifts (b) of the arylacetylide nuclei in [(η²-dppe)(η⁵-

C5Me5)Fe(CtC)-4-(C₆H₄)X][PF₆] complexes (X ) NO₂, CN, CF₃, F, Br, H, Me, OMe, NH₂) vs σ⁺ESPs.

model complexes 1a⁺and 1g⁺featuring the complete coordina-

tion sphere.⁵⁵

As shown in Table 10, the main difference with the

computationally simpler model compounds 1a-H⁺and 1g-H⁺

is that slightly more spin density is located on the metal center

and on the surrounding atoms, in line with the improved

electron-releasing nature of the coordination sphere of 1a⁺and

1g⁺. Also, more spin density is delocalized in the acetylide

spacer in comparison to the simpler model complexes 1a-H⁺

and 1g-H⁺, especially in the latter compound (Table 10).

Notably, in spite of a more marked spin alternation in 1a⁺and

1g⁺, the relative ratios between atomic densities on the

functional aryl group are roughly the same as for 1a-H⁺and

1g-H⁺(Table 9), the discrepancies being slightly more pro-

nounced between 1g⁺and 1g-H⁺. In accordance with previous

findings,¹⁵the largest positive spin density is always located

on the metal center regardless of the conformation adopted, and

then on the â-carbon atom (C₂) of the acetylide ligand.

As shown in Tables 9 and 10, the conformation of the

arylalkynyl ligand has a sizable influence on the spin density

residing on the metal center and on the arylacetylide ligand and

has only a minor influence on the spin distribution in the

phosphane and the cyclopentadienyl ligands. Notably, marked

Figure 10. Plots of the total spin densities for (H₃P)₂(η⁵-C₅H₅)-

changes in the spin distribution within the aryl-alkynyl linker

Fe(CtC)-4-(C₆H₄)X complexes with parallel (left) and perpen-

take place between the nitro compound conformers (1a-H⁺or

dicular (right) conformations of the arylalkynyl ligand. X ) NO₂

1a⁺), but this effect is apparently less pronounced for conformers

of complexes possessing more electron-releasing substituents.

This can be traced back to the different ordering of the frontier

MOs possessing a strong metal character previously pointed out

between compounds with strongly electron-withdrawing sub-

stituents, like 1a-H⁺and compounds with more electron-

releasing substituents.¹⁵As shown in Figure 11, the frontier spin

orbitals with a strong d_yzcharacter are slightly destabilized upon

rotation of the aryl ring from the parallel to the perpendicular

conformation. Conversely, frontier spin orbitals with a strong

d_xzcharacter are slightly stabilized. This leads to a crossing

between the two metal-based frontier R spin orbitals, while the

relative energy ordering of the â spin orbitals is not affected in

1a-H⁺. Thus, the electronic hole remains located in the π_y

(a,b); H (c,d); NH₂(e,f). The contour values are 0.04 [e/bohr³].

(X ) H), 1h-H⁺(X ) OMe), and 1i-H⁺(X ) NH₂) in which

the chelating dppe ligand had been replaced by two PH₃ligands

and the C₅Me₅ligand by C₅H₅.^12,15For these compounds the

spin distribution had been computed in a conformation where

the functional aryl group was roughly coplanar with the C₅H₅

ligand.¹⁵

We now have recalculated the spin densities for 1a-H⁺, 1f-

H⁺, and 1i-H⁺in a perpendicular conformation where the aryl

group has been rotated 90°, and also in both conformations for

the new model compound 1g-H⁺where X ) Me (Table 9).

Importantly, for all Fe(III) model complexes investigated, the

perpendicular conformation appears slightly more stable than

the parallel one (Table 9). Notably, this trend is opposite the

trend previously observed for selected Fe(II) parents.¹²For the

purpose of comparison, spin densities were also derived for

(55) Given the much poorer match obtained with the amino complex

1i-H⁺(see later), we did not compute the spin density for the model complex

1i⁺presenting the exact coordination sphere.

Spin Delocalization in Electron-Rich Iron(III) σ-Acetylides

Organometallics, Vol. 26, No. 4, 2007 887

Table 9. Calculated Spin Densities (e) on Selected Fragments or Atoms for [(η²-dpe)(η⁵-C₅H₅)FeCl]^{+ 27}(dpe )

1,2-diphosphinoethane) and [(PH₃)₂(η⁵-C₅H₅)Fe(CtC-4-C₆H₄X)]⁺Complexes (X ) NO₂, H, NH₂) in Two Conformations (see

Chart 4 for atom numbering)

CtC(C₆H₄)X

∆E^a

R

Fe

C₅H₅

2 PH₃(or dpe)

C₁

C₂

C₃

C₄

C₅

C₆

X

(kJ mol^-1

)

Cl

0.902

0.799

0.640

0.720

0.637

0.578

0.608

0.400

0.436

0.418

-0.023

-0.034

-0.012

-0.023

-0.037

-0.009

-0.023

-0.032

-0.005

-0.019

-0.033^b

-0.026

-0.027

-0.034

-0.014

-0.024

-0.038

-0.023

-0.031

0.154

0.010

0.014

c

CtC(C₆H₄)NO₂

|

-0.054

-0.010

-0.032

0.042

0.030

0.036

0.145

0.089

0.117

0.306

0.236

0.271

0.241

0.222

0.232

0.125

0.148

0.137

-0.027

-0.033

-0.030

0.001

-0.012

-0.006

0.073

0.013

0.085

0.049

0.056

0.088

0.072

0.040

0.045

0.043

0.000

-0.039

-0.020

-0.021

-0.039

-0.030

0.026

0.000

0.104

0.052

0.093

0.137

0.115

0.088

0.084

0.086

14.0

0

d

mean values

CtC(C₆H₅)

0.012

c

-0.007

-0.010

-0.009

0.127

0.106

0.117

e

d

mean values

CtC(C₆H₄)NH₂

c

17.4

0

d

0.050

0.062

0.018

0.022

mean values

^aRelative energy between the two conformations (kJ mol^-1). ^bValue for 1,2-diphosphinoethane (dpe). ^cParallel conformation. ^dPerpendicular conformation.

^eFor this compound, see Computational Details in the Experimental Section.

Table 10. Calculated Spin Densities (mean value in 10³e) for Selected Atoms in [(η⁵-dppe)(η⁵-C₅Me₅)Fe(CtC-4-C₆H₄X)]⁺

and [(PH₃)₂(η⁵-C₅H₅)Fe(CtC-4-C₆H₄X)]⁺Complexes (X ) NO₂: 1a⁺, 1a-H⁺; Me: 1g⁺, 1g-H⁺) (see Chart 4 for

atom numbering)

compd

Fe

P

C₇(dppe)

C₁₁(dppe)

C₁₆(C₅Me₅)

X

∆E^a

b

1a⁺

|

926.8

851.8

799.5

639.8

925.8

772.0

536.2

548.5

-30.2

-28.6

-9.8

4.5

4.2

-0.1

-9.9

-9.2

-8.0

-3.0

-10.4

-8.3

-10.1

-2.2

-0.2

18.5

10.1

18.7

0.1

4.0

11.0

9.2

6.8

0

14.0

0

12.5

0

21.2

0

c

b

1a-H⁺

1g⁺

c

-9.8

b

-30.6

-26.8

-21.2

-9.5

4.8

38.5

0.1

-0.4

c

b

1g-H⁺

c

^aRelative energy between the two conformations (kJ mol^-1). ^bParallel conformation. ^cPerpendicular conformation.

Figure 11. Evolution of the frontier spin-MOs of 1a-H⁺(X ) NO₂) and 1i-H⁺(X ) NH₂) after rotation of the functional aryl ring from

parallel (|) to perpendicular ( ) conformation for (H₃P)₂(η⁵-C₅H₅)Fe(CtC)-4-(C₆H₄)X complexes. The d-metal contribution is shown for

the frontier spin orbitals with strong metal character.

manifold of the acetylide linker in both conformations. In

addition, the lowest unoccupied â spin orbital becomes conju-

gated with the aryl ring in the perpendicular conformation and

the unpaired electron is delocalized on the aryl ligand, while it

was restricted to the acetylide spacer in the parallel conforma-

tion. For strongly electron-releasing substituents, a similar

stabilization/destabilization of the frontier spin orbitals takes

place upon rotation. This time, the crossing occurs for both the

R and â manifolds (Figure 11). Consequently, the electronic

hole “changes” MOs and a quite similar spin distribution results

in both conformers. Finally, for compounds with moderately

electron-releasing or -attracting substituents such as 1f-H⁺, the

SOMO is a mixture of d_yzand d_xzAOs regardless of the

conformation considered (Figure 10), and the rotation of the

aryl ring does not markedly affect the spin delocalization.

Energy differences between 10 and 20 kJ mol^-1are computed

between the two conformations for 1a-H⁺, 1f-H⁺, 1g-H⁺, and

1i-H⁺(Table 9). Note that the steric interactions between

neighboring ligands are not accounted for with such model

compounds, presenting a simplified coordination sphere. No-

tably, significantly smaller energetic differences are found with

the complexes 1a⁺and 1g⁺, presenting the exact coordina-

tion sphere (Table 10). Whereas this might be due to a better

evaluation of the steric interactions taking place between the

888 Organometallics, Vol. 26, No. 4, 2007

Paul et al.

Table 11. Calculated Spin Densities G_C(10³e) for Selected Carbon Atoms in [(PH₃)₂(η⁵-C₅H₅)Fe(CtC-4-C₆H₄X)]⁺(X ) NO₂,

H, Me, NH₂) and in [(η⁵-dppe)(η⁵-C₅Me₅)Fe(CtC-4-C₆H₄X)]⁺Complexes (X ) NO₂, Me) (when available, experimental spin

densities are given for comparison)

CtC-4-(C₆H₄)-

C₃C₄

X

a

C₁

C₂

C₅

-9

C₆

C_X

CtC(C₆H₄)NO₂

b

(F_C)_exp

(F_C)_expratios^c

19

1.00

49

1.00

38

-0.47

-0.46

-20

δ

_Cratios ^c,d

3.81

271

5.53

238

-1.49

-30

1.07

52

1.06

36

(F_C)_DFTfor 1a-H^{+ e}

(F_C)_DFTratios^c

-32

-0.65

-89

-0.61

-34

-0.41

-15

(F_C)_DFTfor 1a^{+ e}

(F_C)_DFTratios^c

-2.31

6.21

-0.88

1.00

-0.40

0.94

CtC(C₆H₅)

b

(F_C)_exp

28

-11

27^f

(F_C)_expratios^c

1.00

72

-0.39

-1.31

-30

0.96^f

2.41

115

1.60

δ

_Cratios^c,d

4.80

232

3.22

-1.98

-6

-0.08

(F_C)_DFTfor 1h-H^{+ e}

(F_C)_DFTratios^c

36

0.50

1.00

-0.42

CtC(C₆H₄)CH₃

b

(F_C)_exp

31

-12

31^g

(F_C)_expratios^c

1.00

80

1.00

38

-0.39

-1.83

-30

1.00^g

2.72

144

1.80

45

δ

_Cratios^c,d

4.98

189

2.36

241

-2.26

13

0.16

-26

-0.68

-1.06

-10

(F_C)_DFTfor 1g-H^{+ e}

(F_C)_DFTratios^c

77

0.96

-65

-1.69

-0.38

-14

-0.37

-0.13

-3

(F_C)_DFTfor 1g^{+ e}

(F_C)_DFTratios^c

6.31

1.00

1.17

-0.09

CtC(C₆H₄)NH₂

b

(F_C)_exp

40

-7

(F_C)_expratios^c

1.00

43

-0.18

δ

_Cratios^c,d

(F_C)_DFTfor 1i-H^{+ e}

(F_C)_DFTratios^c

118

2.74

137

3.19

62

1.44

22

0.51

86

2.00

1.00

^aSee Chart 4 for carbon atom numbering. ^bDetermined from ¹H NMR contact shifts and McConnell equation (in 10³e). ^cRatios relative to the C₄carbon

atom. ^dRatios determined from the uncorrected isotropic ¹³C NMR shifts. ^eMean values between the parallel (|) and perpendicular ( ) conformations of the

arylalkynyl ring (in 10³e). ^fLess precise value determined from “overlapping” para-H shift. ^gLess precise value determined from methyl shift and McConnell

equation (see text).

Chart 5. Related Organic and Organometallic Radicals

arylalkynyl hydrogen atoms and the cyclopentadienyl methyl

groups in the latter set of compounds, the changes in elec-

tronic density on the metal center and within the acetylide linker

in 1a⁺and 1g⁺certainly contribute as well to this decrease.

organometallic complexes were rather scarce until now.^26,56

Again, most of the expected peaks for the various carbon nuclei

for these compounds were detected, except for the nuclei lying

closest to the metal atom such as the R-acetylide carbon (C₁)

and the two ipso carbon atoms of the dppe ligand (C₇and C₁₁).

The failure to detect them most probably results from an

excessive broadening induced by the proximity of the metal

center. A similar statement was previously also made by Ko¨hler

et al. for paramagnetic V(III) arylacetylide complexes such as

8 (Chart 5).²⁶In contrast, for compounds 1i[PF₆] and 1j[PF₆],

Discussion

Assignment of the Paramagnetic NMR Spectra. Most of

the protons of 1a-k[PF₆], 2[PF₆], and 3a,b[PF₆] (Chart 1) have

been detected by NMR and were unambiguously assigned (Table

1). Notably, some previous assignments made for 2[PF₆] and

1f[PF₆] have been presently revised.^21,22We have also shown

here that protons H₃, H₄, H₅, and H₉(Chart 2) exhibit slightly

faster longitudinal (R₁) and transverse (R₂*) relaxation rates than

H₆, H₇, H₈, and H₁₀(Supporting Information). These nuclei relax

faster owing to large contributions of the dipolar mechanism to

the relaxation process, due to their closer proximity to the metal

center.

bearing strongly electron-releasing substituents, many of the ¹³

C

NMR signals pertaining to the aryl acetylide carbon atoms

escaped detection (Table 2). This can be attributed to the larger

spin densities delocalized on this ligand in these complexes,

which in turn induces a faster relaxation of the ¹³C nuclei by

local dipolar effects. However, the â-acetylide carbon atoms

(C₂), which present the largest spin densities of the compounds

after the metal center (Tables 9 and 10), were always observed

as very broad singlets in the 700-800 ppm range with 1a-

h[PF₆] (Table 2).

For the first time, these Fe(III) complexes were also studied

by ¹³C NMR. This is notable, given that studies on related

(56) Ko¨hler, F. H. J. Organomet. Chem. 1974, 64, C27-C28.

Spin Delocalization in Electron-Rich Iron(III) σ-Acetylides

Organometallics, Vol. 26, No. 4, 2007 889

Table 12. Calculated Spin Densities for H Atoms in

[(PH₃)₂(η⁵-C₅H₅)Fe(CtC-4-C₆H₄X)]⁺(X ) NO₂, H, Me,

NH₂) and in [(η⁵-dppe)(η⁵-C₅Me₅)Fe(CtC-4-C₆H₄X)]⁺

Complexes (X ) NO₂, Me) (when available, experimental

spin densities are given for comparison)

groups (compare for instance the spectra of 1f[PF₆] with those

of 1g[PF₆] or 3a,b[PF₆]),⁵⁹while substitution of the para (H_X)

or meta (H₂) protons in 1f[PF₆] for fluorine or trifluoromethyl

substituents produces ¹⁹F NMR shifts similar to those observed

with fluorinated π-radicals.^40,60,61

CtC-4-(C₆H₄)-

X

The |a_H| values derived for the nuclei of the arylacetylide

ligand (Table 4) are below those usually observed with purely

organic aryl-centered radicals like the phenylalkynyl radical

anion 5^-(Chart 5 and Table 13).⁶²Also, the |a_F| hyperfine

constants derived for 1d[PF₆] and m-1d[PF₆] (Table 3) are

roughly one-third of those previously found for fluorinated

phenoxy radicals.⁶³Given that the unpaired electron density in

1a-j[PF₆] is mostly localized on the metal center,¹⁵these Fe-

(III) complexes should preferably be compared with phenyla-

lkynyl or γ-phenylpropargyl radicals like 6 and 7 (Chart 5), for

which lower |a_H| values were reported (Table 13).⁶⁴Thus, the

unpaired electron is approximately 4 time less delocalized on

the aryl ring in 1a-j[PF₆] than in 7.

a

H₁

H₂

H_X

CtC(C₆H₄)NO₂

-0.64

1.00

-3.7

1.00

b

(F_H)_exp

0.30

-0.47

-0.57

1.8

-0.49

1.3

(F_H)_expratios^c

δ_iratios^c,d

(F_H)_DFTfor 1a-H^{+ e}

(F_H)_DFTratios^c

(F_H)_DFTfor 1a^{+ e}

(F_H)_DFTratios^c

-2.7

1.00

-0.46

CtC(C₆H₅)

-0.93

1.00

-5.8

1.00

b

(10³F_H)_exp

0.36

-0.39

-0.45

1.9

-0.88

0.95

1.00

-8.5

1.46

(F_H)_expratios^c

δ_iratios^c,d

(F_H)_DFTfor 1h-H^{+ e}

(F_H)_DFTratios^c

-0.33

Also, in accordance with theoretical predictions, larger

hyperfine constants are observed for the ortho (H₁) and para

(H_X) protons than for meta (H₂) protons. This makes sense, since

the carbon atoms C₄and C₆bearing these protons are conjugated

with the metal center and consequently drain more spin density

than do the meta carbon atoms (C₅) due to the occurrence of

Fermi delocalization.¹⁵Along similar lines, the very slight

difference in paramagnetic shift between the inequivalent H₁

or C₄nuclei in m-1d[PF₆] can be understood considering that

the fluorine substituent interacts in a π-fashion with an MO

“orthogonal” to that containing the unpaired electron in this

compound, thus inducing only a weak electronic perturbation

on the shifts of these nuclei.

CtC(C₆H₄)CH₃

-1.04

1.00

-6.4

1.00

b

(10³F_H)_exp

0.38

-0.37

-0.42

1.3

-0.20

1.3

1.18

-1.13

-1.22

6.6

-1.03

1.8

(F_H)_expratios^c

δ_iratios^c,d

(F_H)_DFTfor 1g-H^{+ e}

(F_H)_DFTratios^c

(F_H)_DFTfor 1g^{+ e}

(F_H)_DFTratios^c

-2.6

1.00

-0.50

-0.69

CtC(C₆H₄)NH₂

-1.37

1.00

-4.1

1.00

b

(10³F_H)_exp

0. 25

-0.18

-0.20

-2.1

-0.12

0.09

0.08

-4.0

0.98

(F_H)_expratios^c

δ_iratios^c,d

(F_H)_DFTfor 1i-H^{+ e}

(F_H)_DFTratios^c

-0.51

^aSee Chart 2 for proton numbering. ^bDetermined from ¹H NMR contact

shifts (in 10³e Å^-3). ^cRatios relative to the value found for proton H₁.

^dRatios determined from the uncorrected isotropic ¹H NMR shifts. ^eMean

values between the parallel (|) and perpendicular ( ) conformations of the

arylalkynyl ring (in 10³e).

In comparison to the arylacetylide ligand, the spin delocal-

ization in the η⁵-C₅Me₅and η²-dppe ligands is lower. In 1a-

1

j[PF₆], the overall H NMR shift to high field stated for the

methyl groups reveals a large and negative spin density on the

π manifold of the five-membered ring, in accord with DFT

computations (Tables 9 and 10). In contrast, a positive spin

density might be present within the σ-system of the C₅ring

according to the ¹³C NMR shifts. The NMR data would therefore

be in line with a σ-type spin delocalization mechanism.⁶⁵Given

the relative constancy of the geometric parameters, the pseudo-

contact contribution to the ¹H NMR shifts should not vary much

between 1a-j[PF₆] and 2[PF₆]. Consequently, considering that

For related reasons, the detection of the ³¹P signals of the

dppe ligand for 1a[PF₆] is quite remarkable as well. Indeed,

only in rare instances were ³¹P NMR peaks reported for

paramagnetic complexes featuring a phosphane ligand directly

coordinated to the metal center.⁵⁷The large upfield shift

observed for 1a[PF₆] is sensible, since the isotropic hyperfine

coupling constant (|a_P|) deduced from this isotropic shift (ca.

14 G) by neglecting the pseudocontact contribution is very

similar to the mean diagonal value of the anisotropic hyperfine

coupling tensor previously reported for [(η²-dppm)(η⁵-C₅Me₅)-

FeCtC-Ph][PF₆] (|A_P|_xx≈ |A_P|_yy≈ |A_P|_zz≈ 14-15 G).

Actually, this Fe(III) complex is the dppe analogue of 1f[PF₆]

and its |A_P| values were measured by ESR.⁵⁸

1

the changes in the H NMR and ¹³C NMR shifts of the η⁵-C₅-

Me₅ligand mostly originate from changes in the contact

contribution, the isotropic NMR shifts tend to indicate that the

spin density on the cyclopentadienyl ligand has slightly increased

upon proceeding from 1a-j[PF₆] to 2[PF₆]. In line with such a

hypothesis, the half-widths of the peak of the methyl group from

the cyclopentadienyl ligand observed for 2[PF₆] are significantly

Spin Delocalization in Fe(III) Radicals. Regarding the spin

distribution in these compounds, the ¹H NMR contact shifts in

1a-j[PF₆] are clearly diagnostic of a sizable delocalization of

the unpaired electron in a π MO of the arylacetylide ligand.^25,26

As expected from polarization effects operative in an alternant

π manifold (Scheme 3), opposite shifts are observed for protons

1

broader (ca. 2115 Hz) in comparison to the corresponding H

NMR signals for 1a-j[PF₆] or 4[PF₆] (e775 Hz). This also

suggests that the arylacetylide ligand competes more efficiently

(60) Eaton, D. R.; Josey, A. D.; Benson, R. E.; Phillips, W. D.; Cairns,

T. L. J. Am. Chem. Soc. 1962, 84, 4100-4106.

(61) Icli, S.; Kreilick, R. W. J. Phys. Chem. 1971, 75, 3462-3465.

(62) (a) Evans, A. G.; Evans, J. C.; Phelan, T. J. J. Chem. Soc., Perkin

Trans. 2 1974, 1216-1219. (b) Evans, A. G.; Evans, J. C.; Emes, P. J.;

Phelan, T. J. J. Chem. Soc. B 1971, 315-318.

(63) Espersen, W. G.; Kreilick, R. W. Mol. Phys. 1969, 6, 407-416.

(64) (a) Coleman, J. S.; Hudson, A.; Root, K. D. J.; Walton, D. R. M.

Chem. Phys. Lett. 1971, 11, 300-301. (b) Kochi, J. K.; Krusic, P. J. J. Am.

Chem. Soc. 1970, 92, 4110-4114.

(65) Rettig, M. F. In NMR of Paramagnetic Molecules; La Mar, G. N.,

Horrocks, D., Holm, R., Eds.; Academic Press: New York, 1973; pp 217-

242.

1

on adjacent carbon atoms. Also, H NMR shifts in opposite

directions are observed upon replacement of H₁or H₂by methyl

(57) Mink, L. M.; Polam, J. R.; Christiensen, K. A.; Bruck, M. A.;

Walker, F. A. J. Am. Chem. Soc. 1995, 117, 9329-9339.

(58) Connelly, N. G.; Gamasa, M. P.; Gimeno, J.; Lapinte, C.; Lastra,

E.; Maher, J. P.; Le Narvor, N.; Rieger, A. L.; Rieger, P. H. J. Chem. Soc.,

Dalton Trans. 1993, 2575-2578.

(59) (a) Lazdins, D.; Karplus, M. J. Am. Chem. Soc. 1965, 24, 920-

921. (b) Forman, A.; Murrell, J. N.; Orgel, L. E. J. Chem. Phys. 1959, 31,

1129, and references therein.

890 Organometallics, Vol. 26, No. 4, 2007

Paul et al.

with the η⁵-C₅Me₅ligand in 1a-j[PF₆] than does the chloride

ligand in 2[PF₆] for delocalizing the unpaired electron. Thus, it

is the arylacetylide linker and not the C₅Me₅ligand that most

strongly contributes to the delocalization of the unpaired electron

in 1a-j[PF₆].²²Other experimental facts in agreement with this

proposal are the much larger shifts observed for the protons of

the para-phenyl substituent shifts in 4[PF₆] in comparison to

those of the aromatic dppe signals.

combination with eq 4.^17,72As shown in Table 11, the spin

densities determined experimentally using McConnell’s expres-

sion were in general smaller than those computed by DFT. This

is not surprising considering that DFT has a tendency to

overestimate spin delocalization.⁷³For comparison purposes,

their relative ratios were therefore also given in Tables 11 and

12.

Regarding the diphosphine ligand, the ³¹P NMR shift to high

field observed for 1a[PF₆] clearly reveals the presence of a

negative spin density on the phosphorus nuclei, which most

likely results from polarization effects, according to the DFT

calculations (Tables 9 and 10). Due to their low magnitude,

the ¹H NMR contact shifts for protons of the dppe ligand should

be interpreted more cautiously, especially without any accurate

determination of the pseudocontact shifts.⁶⁶The DFT calcula-

tions indicate that only a slight amount of negative spin density

is directly delocalized from the phosphorus atoms to the phenyl

rings by Fermi contact. Although the π manifold certainly

contributes to this process, part of the contact shift also originate

from unpaired spin density in the σ manifold, since the typical

a_H) (2µ₀/3)(pγ_Hg_eµ_B)F_H

(12)

In line with the weak energetic differences (below 22 kJ

mol^-1) computed between the parallel and perpendicular

conformations for 1a-H⁺, 1f-H⁺, 1g-H⁺, and 1j-H⁺(Tables 9

and 10), facile rotation of the aryl ring can be anticipated at

ambient temperature. Accordingly, variable-temperature NMR

experiments performed on selected Fe(II) samples suggest that

the arylalkynyl ligand is rotating freely in solution down to 200

K.³⁰In order to compare computed spin densities with spin

densities deduced from NMR measurements in solution, the

average between the theoretical spin densities calculated for the

two conformers was taken into consideration (Tables 11 and

12).

1

alternation between the H NMR shifts of vicinal protons is

not observed for the aromatic protons H₃-H₅and H₆-H₈(Table

1).⁶⁸Such behavior is apparently common for arylphosphine

ligands bound to paramagnetic metal centers.⁶⁹

Regarding spin densities on the protons of the arylacetylide

ligand, it can be stated that ratios between empiric values

obtained for the spin densities compare remarkably well with

theoretical (DFT) values computed in the case of complexes

bearing an electron-withdrawing substituent like 1a[PF₆], and

this is regardless of the accuracy of the model compound used

in the calculations (1a-H⁺or 1a⁺). The match is excellent,

especially when considering the approximations made during

their experimental determination along with the inherent preci-

sion of DFT computations where no solvent nor counterion was

considered (Tables 11 and 12).⁷⁴In contrast, for the compound

1j-H⁺with a strongly electron-releasing substituent, the match

is much poorer. Several explanations can be put forward to

rationalize this discrepancy. First, it was noticed earlier that

when the substituent becomes more electron-releasing, larger

differences between computed spin distributions are found for

the arylacetylide ligand between model complexes featuring the

exact coordination sphere and simpler model complexes. Thus,

1j-H⁺might be less adapted than a more accurate model

compound like 1j⁺for computing the spin densities in these

cases. Then, for complexes bearing electron-releasing substit-

uents, the non-negligible influence of local dipolar effects on

the isotropic shifts has also been previously put forward (Figure

8). The importance of such effects on relaxation rates has been

discussed by Doddrell and co-workers.^49,50However, estimation

of their influence on the isotropic shifts is outside the scope of

this work, since it is far from being trivial.⁷⁵Thus, given that

these effects are not accounted for in the classical treatment

used here, an increasingly wrong contact shift is possibly

determined from experiment by our approach (eq 3) for

compounds with strongly electron-withdrawing substituents.

Similar to eq 5, an equation can be written for the fluorine

¹⁹F isotropic constants (eq 11). In that case, due to the presence

of a p-type AO on the fluorine, polarization effects must be

accounted for (eq 11).¹⁷Using eq 11 in combination with eq 4

F

and the carbon spin densities of 1f[PF₆],⁷⁰along with the (Q_app

)

F

values previously proposed by Eaton and co-workers ((Q_app

)

) 73.8 G/m-F, 52.6 G/p-F, 38.4 G/p-CF₃),^60,71the ¹⁹F NMR

contact shifts were computed for 1c[PF₆], 1d[PF₆], and m-1d[PF₆]

and amount to 94.7, 128.0, and -75.5 ppm, respectively. These

shifts are clearly different from the shifts found when the

isotropic shift is simply corrected for the metal-centered

pseudocontact contribution (45.5, 132.5, and -21.8 ppm), but

have the correct sign. Thus, if the constants used in eq 11 are

indeed appropriate for these compounds,⁶³this constitutes

another indication of the existence of additional non-negligible

ligand-centered pseudocontact contributions (δ_pc)^Lto the ¹⁹F

NMR isotropic shifts. Such effects were previously neglected

by Eaton and co-workers.

a_F) (Q_CF)^C(F_C)^π+ [(S_F)^F+ (Q_FC)^F](F_F)^π) {(Q_CF)^F+

Ap_CF[(S_F)^F+ (Q_FC)^F]}(F_C)^π) (Q_app)^F(F_C)^π(11)

Comparison between Spin Densities Derived from the

Experiment (NMR) and Theory (DFT). Spin densities for

carbon nuclei were derived from ¹H NMR contact shifts using

the McConnell equation (eq 5) and for protons using eq 12 in

(66) As pointed out by one referee, for compounds exhibiting a slight

rhombic distortion of the g tensor, equations making use of the susceptibility

anisotropy values should have been used instead of eq 3.¹⁷However, in

order to obtain very accurate values of the metal-centered pseudocontact

shifts, a weighted average of all different conformations of the compounds

should also have been considered.⁶⁷This proved clearly unfeasible with

such labile compounds, especially considering the large number of degrees

of freedom of the dppe ligand in these complexes (see for instance Scheme

2).

(70) It was often proposed that hydrogen for fluorine substitution does

not affect the π-spin density on the adjacent carbon atom.^39,71

(71) Eaton, D. R.; Josey, A. D.; Phillips, W. D.; Benson, R. E. Mol.

Phys. 1962, 5, 407-416.

(72) In this expression, F_His the spin density residing in the s orbital of

the observed H nucleus. Other symbols have their usual meaning, which

was defined previously.³⁴

(73) Ciofini, I.; Illas, F.; Adamo, C. J. Chem. Phys. 2004, 120, 3811-

3816, and references therein.

(74) Adamo, C; Subra, R.; Di Matteo, A.; Barone, V. J. Chem. Phys.

1998, 109, 10244-10254.

(75) Gottlieb, H. P. W.; Barfield, M.; Doddrell, D. M. J. Chem. Phys.

1977, 67, 3785-3794.

(67) Horrocks, W. D., Jr.; Greenberg, E. Inorg. Chem. 1971, 10, 2190-

2194.

(68) Horrocks, W. D., Jr.; Taylor, R. C.; La, Mar, G. N. J. Phys. Chem.

1964, 86, 3031-3038.

(69) (a) Doddrell, D. M.; Roberts, J. D. J. Am. Chem. Soc. 1970, 92,

6839-6844. (b) Moroshima, I.; Yonezawa, T.; Goto, K. J. Am. Chem. Soc.

1970, 92, 6651-6653.

Spin Delocalization in Electron-Rich Iron(III) σ-Acetylides

Organometallics, Vol. 26, No. 4, 2007 891

Scheme 3. Qualitative Spin Distribution on the

Arylacetylide Ligands Deduced from the H NMR Contact

unpaired spin density will spread more and more on the

arylalkynyl ligand in proportion to the electron-releasing

capability of the para substituent. Further, the observed linear

correlations with σ⁺ESPs (Figure 9) suggest that it results from

a purely electronic substituent effect, depending largely on the

π-interaction of the X substituent with the aryl ring. On the

basis of these correlations, it seems clear that NMR provides a

simple way to check the nature of an unknown para substituent

in similar compounds. However, this linear dependence appar-

ently breaks down for complexes bearing very electron-releasing

substituents such as 1j[PF₆], an observation that might again

be related to the existence of strong local dipolar effects.

Note that usual ESP sets are traditionally considered as

reflecting substituent-induced changes in charge distribution

within the functional aryl group, not changes in spin distribu-

tion.^54,77Although we could not find any proportionality between

the computed Hirschfeld charges and spin densities for any Fe-

(III) model complex presently investigated, a rough cor-

respondence between the substituent-induced changes for these

two data sets along the 1a-H⁺/1j-H⁺series was apparent for

C₁, C₂, and C₃(Supporting Information).¹²In line with this

observation, approximate linear correlations between the com-

puted spin densities and the σ⁺ESPs were also found (Figure

12) for these nuclei. This can presently explain the linear free

energy relationships (LFER) observed with the ¹H NMR

isotropic shifts. However, the spin and charge variations induced

by the substituents on C₄-C₆are much smaller. In regard to

the accuracy of the DFT method, further rationalization of the

observed LFERs on the basis of the present calculations is

therefore quite tentative (Figure 9a).

1

Shifts

Table 13. Hydrogen Hyperfine Coupling Constants (G) with

Aryl Protons Determined for Various Organic Aryl-Based

Radicals at Different Temperatures

compd

T (K)

|a_H|_ortho

0.16

2.22

2.55

|a_H|_meta

3.16

0.81

0.82

|a_H|_para

7.49

2.16

2.55

5[K]

6

7

193

273-263

<273

This, in turn also certainly contributes to overestimating the real

spin density on carbon nuclei such as C₄.

The good agreement observed between computed DFT values

and carbon spin densities ratios deduced using the McConnell

equation in most compounds among 1a-j[PF₆] (Table 11) is

also worth mentioning.⁷⁴This indicates that a π-σ polarization

mechanism dominates the NMR contact shifts of the arylacetyl-

ide protons and, in accordance with calculations, confirms that

the unpaired electron is mostly delocalized on the π manifold

of the arylacetylide ligand. Indeed, such a good correspondence

should not be obtained if a sizable direct spin delocalization

had taken place via the σ manifold. Regarding the arylacetylide

nuclei, we also show (Table 12) that fair relative estimates of

the atomic spin densities can already be obtained from ratios

between uncorrected isotropic ¹H NMR shifts. This is however

clearly not the case for the corresponding ¹³C NMR isotropic

shifts (Table 11). Due to the existence of p-type AOs on carbon

nuclei, these shifts constitute far less reliable indicators of atomic

spin densities due to the contribution of polarization effects

along with non-negligible local dipolar effects.

Ko¨hler and co-workers have previously shown that the

conformation adopted by the aryl ring in the arylacetylide ligand

can have a strong influence on the spin delocalization within

this ligand (Figure 10).⁵⁶Likewise with 1a-i⁺complexes, DFT

calculations suggest that the changes in spin distribution between

the two conformations will be especially marked for compounds

bearing strongly electron-withdrawing substituents such as 1a⁺

(Scheme 4). In other cases, the electronic relaxation after rotation

results in a quite analogous spin distribution in parallel and

perpendicular conformations. Thus, at least in compounds with

electron-withdrawing substituents like 1a[PF₆], the rotation of

the phenyl ring does slightly modulate the spin density on the

metal center and on the C₅Me₅ring. For these compounds, the

internal correlation time (τ_i) associated with this motion might

therefore influence the correlation time (τ_c) corresponding to

the dipolar relaxation of nuclei on other ligands (eq 9).

Substituent Influence on Spin Delocalization. To our

knowledge, while substituent effects on the NMR shifts have

previously been investigated for diamagnetic organic alkynes,⁷⁶

no such investigations have been conducted for paramagnetic

alkynyl complexes. For 1a-h[PF₆], both DFT calculations on

model compounds and ¹H NMR shifts clearly confirm that the

Linewidth Data and Self-Exchange Rates. A maximum

value of the electronic correlation time (τ_e< 8.5 × 10^-11s)

1

was derived above from the linewidths of the H NMR peaks

of 1a[PF₆] and 1g[PF₆] according to Bloembergen-Solomon-

like expressions (eqs 7a-d).⁷⁸Now, considering that the half-

widths of the ortho-acetylide protons (H₁) are solely determined

by the contact relaxation (see above), these values in 1a[PF₆]

and 1g[PF₆] can be further precised, although the latter complex

constitutes obviously a limiting case for such an approach, due

to the neglect of any local dipolar effect on the proton spin

relaxation (Figure 8). Values around 2.2 × 10^-11and 3.5 ×

10^-11s can thus be derived for τ_e, respectively. Considering

the non-negligible energy differences found by DFT between

the two extreme conformations of the aryl ring (Table 10), we

can suppose that for H₁and H₂the rotation rate of the aryl ring

around the acetylide axis is much slower than the electron spin

relaxation (τ_i. τ_s). In the absence of other internal motion

(see eq 9), the electronic correlation time presently determined

therefore roughly corresponds to the spin correlation time (τ_s)

under the assumption of equal longitudinal and transverse

relaxation times (τ_s≈ τ_1s≈ τ_2s).¹⁷Thus τ_ewould roughly

correspond to the relaxation time of the unpaired electron (τ_s),

for which values around (2-4) × 10^-11s can be proposed.

Actually, these estimates are somewhat higher than most τ_s

values that were previously reported for low-spin Fe(III)

complexes (closer to 10^-12s).¹⁷However, to our knowledge,

essentially purely inorganic Fe(III) complexes were investigated

in terms of τ_suntil now, not organometallic ones such as 1a-

j[PF₆]. Moreover, the data tend to indicate that τ_svalues slightly

(77) (a) Jiang, X.-K. Acc. Chem. Res. 1997, 30, 283-289. (b) Shorter,

J. Chem. Z. 1985, 19, 197-208. (c) Jaffe´, H. H. Chem. ReV. 1953, 53,

191-261.

(78) (a) Solomon, I. Phys. ReV. 1955, 99, 559-566. (b) Bloembergen,

N. J. Chem. Phys. 1957, 27, 572-573. (c) Bloembergen, N.; Purcell, E.

M.; Pound, R. V. Phys. ReV. 1948, 73, 679-715.

(76) Rubin, M.; Trofimov, A.; Gevorgyan, V. J. Am. Chem. Soc. 2005,

127, 10243-10249, and references therein.

892 Organometallics, Vol. 26, No. 4, 2007

Paul et al.

Scheme 4. Changes in SOMO in the Two Conformations Resulting from a 90° Rotation of the Aryl Ring in 1a-j[PF₆] for

Strongly Electron-Withdrawing Substituents (A) and in Other Cases (B)

increase when the substituent becomes more electron-releasing,

in accordance with ESR measurements.¹⁵

of these electron-transfer reactions.^51,80Such low reorganization

energies were expected from previous investigations on 1a-

j[PF₆].^12,14Actually, these values compare quite well with values

around 4000-5000 cm^-1derived for related dinuclear com-

plexes in the mixed-valent Fe(II)/Fe(III) state from the inter-

valence charge-transfer (IVCT) bands.^10aThis suggests that the

electronic reorganization taking place during the electron-transfer

event involves primarily the “(η²-dppe)(η⁵-C₅Me₅)FeCtC-”

fragment.

Self-exchange rates for several redox couples were also

derived from half-widths of Fe(II)/Fe(III) mixtures at various

temperatures. The rates found range from ca. 10 × 10⁷to 25.8

× 10⁷M^-1s^-1at 25 °C. These rates are faster for 1a-j[PF₆]

than for 2[PF₆], and among 1a-j[PF₆] appear faster for

complexes with electron-withdrawing substituents. Activation

energies around 9-12 kJ mol^-1were derived for these processes

(Table 8). The close similarity between these values suggests

that the changes in self-exchange rates among 1a-j[PF₆] are

not solely determined by changes in the corresponding activation

energies, but that they will also strongly depend on differences

in the Arrhenius prefactors. Unfortunately, given that Arrhenius

prefactors are obtained from Eyring plots with very large

uncertainties, no further rationalization of the substituent influ-

ence on the self-exchange rates was attempted. In comparison

to other families of redox couples, the self-exchange rates found

for 1a-j[PF₆] or 2[PF₆] are slightly faster than these previously

determined for tris-diimine cationic iron complexes such as [Fe-

Conclusions

Several paramagnetic Fe(III) mononuclear arylacetylide com-

plexes of formula (η²-dppe)(η⁵-C₅Me₅)Fe-CtC-Ar (Ar )

functional aryl group) were studied by multinuclear NMR. The

signals of most of the nuclei could be detected, and a complete

assignment of the various ¹H, ¹³C, and in some cases ¹⁹F signals

was proposed for the corresponding NMR spectra. It clearly

appears from this work that the arylacetylide linker and not the

permethylated cyclopentadienyl ligand quite strongly contributes

to the delocalization of the unpaired electron in 1a-j[PF₆], in

contrast to what was previously found for 2[PF₆]. In connection

with this observation, we show here that the ¹H NMR shifts of

the ortho (H₁) and meta (H₂) arylacetylide protons can be

interpreted in terms of dominant contact contribution and bring

decisive information on the spin distribution within the aryl ring

for most compounds investigated using fairly simple treatments.

Estimates of the proton (a_H) and fluorine (a_F) hyperfine coupling

constants were therefore derived for nuclei of the arylacetylide

linker. These data are diagnostic of an unpaired electron

dominantly centered on the metal center and partly residing in

a π MO on the arylacetylide ligand. All these statements are

also well substantiated by DFT calculations on computationally

simpler model compounds. Sensible estimates of the electronic

relaxation times around 2 × 10^-11s could be derived from the

(bipy)₃]³⁺/[Fe(bipy)₃]^{2+ 52}

rates reported for anionic hexacyano complexes such as [Fe-

(CN)₆]^3-/[Fe(CN)₆]^{4- 79}

Considering an outer-sphere process

,

but much faster than self-exchange

.

with negligible electronic coupling, reorganization energies

around 4000 cm^-1can be deduced from the activation energies

1

half-widths of the various H NMR peaks of 1a-j[PF₆] at

ambient temperatures. An important finding of this work is also

1

that quite sizable contributions to the H NMR isotropic shifts

might come from local dipolar effects due to the relatively large

(79) Shporer, M.; Ron, G.; Loewenstein, A.; Navon, G. Inorg. Chem.

1965, 4, 361-364.

(80) (a) Brunschwig, B. S.; Creutz, C.; Sutin, N. Chem. Soc. ReV. 2002,

31, 168-184. (b) Nelsen, S. F.; Blackstock, S. C.; Kim, Y. J. Am. Chem.

Soc. 1987, 109, 677-682. (c) Creutz, C.; Newton, M. D.; Sutin, N. J.

Photochem. Photobiol. A 1994, 82, 47-59.

Figure 12. Plot of the computed spin density for selected carbon

atoms of the arylacetylide nuclei in [(PH₃)₂(η⁵-C₅H₅)Fe(CtC)-4-

(C₆H₄)X]⁺complexes (X ) NO₂, CN, Br, H, Me, OMe, NH₂) vs

+

σ

ESPs.

Spin Delocalization in Electron-Rich Iron(III) σ-Acetylides

Organometallics, Vol. 26, No. 4, 2007 893

Synthesis of the Mononuclear Fe(II) Alkynyl Complex (η²-

dppe)(η⁵-C₅Me₅)Fe-CtC-3,5-(C₆H₃(CH₃)₂) (3a). (η²-dppe)(η⁵-

C₅Me₅)Fe(Cl) (2, 1.000 g, 1.60 mmol), NH₄PF₆(0.315 g, 1.93

mmol), and (3,5-dimethylphenyl)acetylene (0.360 g, 2.77 mmol)

were suspended in 50 mL of methanol, and the mixture was stirred

12 h at 25 °C. After concentration to 10 mL and decantation, the

orange solid was filtrated, washed with methanol (5 mL), and

extracted with 50 mL of dichloromethane. Concentration of the

extract (2 mL) and precipitation by excess diethyl ether (20 mL)

allowed the isolation of [(η²-dppe)(η⁵-C₅Me₅)FedCdCH{3,5-

(C₆H₃)(CH₃)₂}][PF₆] (7a[PF₆]) as an air-sensitive orange solid

(1.020 g, 1.182 mmol, 74%). This vinylidene complex (0.800 g,

0.927 mmol) was then stirred for 3 h in THF in the presence of

excess potassium tert-butoxide (0.157 g, 1.391 mmol). After

removal of the solvent and extraction with toluene, concentration

of the extract to dryness, and subsequent washings with n-pentane,

the desired orange (η²-dppe)(η⁵-C₅Me₅)Fe[CtC-3,5-(C₆H₃)(CH₃)₂]

complex 3a was obtained (0.430 g, 0.599 mmol, 64%). X-ray-

quality crystals of 3a were grown upon cooling the washings to

-30 °C (see Table 14 for selected bond lengths and angles). Total

yield: 47%. Color: red-orange. MS (LSI⁺, m-NBA): m/z 718.2585

([M]⁺, 30%), m/z calc for [C₄₆H₄₈P₂⁵⁶Fe]⁺) 718.2581. Anal. Calcd

for C₄₆H₄₈P₂Fe₁: C, 76.88; H, 6.73. Found: C, 76.84; H, 6.84.

FT-IR (ν, KBr/CH₂Cl₂, cm^-1): 2052/2048 (s, CtC). Raman (neat,

ν, cm^-1): 2053 (vs, CtC). ³¹P{¹H} NMR (C₆D₆, 81 MHz, δ in

ppm): 101.7 (s, dppe). ¹H NMR (C₆D₆, 200 MHz, δ in ppm): 8.06

(m, 4H, H_{ortho/Ph1/dppe}); 7.29-6.94 (m, 18H, H_Ar/dppe+ H_oMes); 6.67

(s, 1H, H_pMes); 2.70 (m, 2H, CH_2dppe); 2.19 (s, 6H, 2CH₃); 1.85 (m,

2H, CH_2dppe); 1.57 (s, 15H, C₅(CH₃)₅). ¹³C{¹H} NMR (C₆D₆, 50

spin density present on the arylacetylide ligand in compounds

bearing electron-releasing substituents. Finally, estimates of the

self-exchange rates in the (10-26) × 10⁷M^-1s^-1range were

derived for these complexes from line-broadening studies of

Fe(II)/Fe(III) mixtures. These self-exchange rates are slightly

substituent dependent and are apparently faster for the com-

pounds with electron-withdrawing groups. Reorganization ener-

gies around 4000 cm^-1could be obtained for the associated

electron-transfer process, which compare quite well with previ-

ous estimates of similar quantities in related dinuclear mixed-

valent complexes. In conclusion, the present contribution clearly

shows that multinuclear NMR constitutes a very powerful tool

to study paramagnetic electron-rich Fe(III) acetylide complexes

with S ) 1/2 such as 1a-g[PF₆] in solution. Provided the

appended substituent is not too strongly electron-releasing, this

technique allows mapping the spin distribution in such com-

pounds in a quite straightforward way.

Experimental Section

General Data. All manipulations were carried out under inert

atmospheres. Solvents or reagents were used as follows: Et₂O and

n-pentane, distilled from Na/benzophenone; CH₂Cl₂, distilled from

CaH₂and purged with argon; HN(ⁱPr)₂, distilled from KOH and

purged with argon; aryl bromides (Acros, >99%), opened/stored

under Ar. The [(η⁵-C₅H₅)₂Fe][PF₆] ferricinium salt was prepared

by previously published procedures.⁸¹Transmittance-FTIR spectra

were recorded using a Bruker IFS28 spectrometer (400-4000

cm^-1). Raman spectra of the solid samples were obtained by diffuse

scattering on the same apparatus and recorded in the 100-3300

cm^-1range (Stokes emission) with a laser excitation source at 1064

nm (25 mW) and a quartz separator with a FRA 106 detector. Near-

infrared (NIR) spectra were recorded using a Bruker IFS28

spectrometer, using a Nernst Globar source and a KBr separator

with a DTGS detector (400-7500 cm^-1), a tungsten source and a

quartz separator with a Peltier-effect detector (5200-12500 cm^-1),

or on a Cary 5 spectrometer (4000-12500 cm^-1). UV-visible

spectra were recorded on an Uvikon XL spectrometer (250-12500

cm^-1). All NMR experiments were made on a Bruker Avance 500

operating at 500.15 MHz for ¹H, 125.769 MHz for ¹³C, and 201.877

MHz for ³¹P, with a 5 mm broadband probe equipped with a

z-gradient coil. More details about the NMR experiments are

provided as Supporting Information. ESR spectra were recorded

on a Bruker EMX-8/2.7 (X-band) spectrometer. Cyclic voltammo-

grams were recorded using a EG&G potentiostat (M.263) on

platinum electrodes referenced to an SCE electrode and were

calibrated with the Fc/Fc⁺couple taken at 0.46 V in CH₂Cl₂.⁸¹MS

analyses were performed at the “Centre Re´gional de Mesures

Physiques de l’Ouest” (CRMPO, University of Rennes) on a high-

resolution MS/MS ZABSpec TOF micromass spectrometer. El-

emental analyses were performed at the CRMPO or at the Centre

for Microanalyses of the CNRS at Lyon-Solaise, France.

MHz, δ in ppm): 141.0-127.5 (m, 8C_Ar/dppe+ 2 CH_Mes); 137.2 (s,

2

C

_quat./Mes); 131.6 (s, C_quat./Mes); 136.0 (t, J_CP) 40 Hz, Fe-CtC);

125.9 (s, C-H_Mes); 121.3 (s, Fe-CtC); 88.3 (s, C₅(CH₃)₅); 31.4 (m,

CH_2/dppe); 21.2 (s, CH₃); 11.1 (s, C₅(CH₃)₅). CV: E₀(∆E_p, i_pa/i_pc)

-0.19 V (0.07, 1) vs SCE. UV-vis (CH₂Cl₂): λ_max(ꢀ/10³dm³M^-1

cm^-1) 246 (sh, 28.0); 284 (sh, 13.1); 346 (12.8).

Selected Data for [(η²-dppe)(η⁵-C₅Me₅)FedCdCH{3,5-(C₆H₃)-

(CH₃)₂}][PF₆] (7a[PF₆]). FT-IR (ν, KBr, cm^-1): 1618 (s, Fed

CdC). ³¹P{¹H } NMR (CDCl₃, 81 MHz, δ in ppm): 89.0 (s, dppe);

143.1 (septuplet, J_PF) 712 Hz, PF₆^-). H NMR (δ, CDCl₃, 200

MHz): 7.70-7.20 (m, 16H, H_Ar/dppe); 7.15 (m, 4H, H_Ar/dppe); 6.68

1

4

(s, 1H, H_Mes); 6.05 (s, 2H, H_Mes); 5.09 (t, J_HP) 4 Hz, 1H, Fed

CdC(Ar)H); 3.08 (m, 2H, CH_2dppe); 2.52 (m, 2H, CH_2dppe), 2.04

(s, 6H, CH₃); 1.59 (s, 15H, C₅(CH₃)₅). ¹³C NMR (CDCl₃, 50 MHz,

2

δ in ppm): 362.7 (m, J_CP) 34 Hz, J_CH) 6.1 Hz, FedCdC);

4

1

138.0 (m, J_CH) 5 Hz, J_CH) 152 Hz, C-CH_3/Mes); 133.8-128.6

3

1

(m, 12C_Ar); 126.4 (m, J_CP) 5 Hz, J_CH) 152 Hz, FedCdC);

1

124.8 (d, J_CH) 157 Hz, CH_Mes); 100.5 (s, C₅(CH₃)₅); 29.4 (m,

1

CH_2/dppe); 21.4 (q, J_CH) 126 Hz, CH_3/Mes); 10.5 (q, J_CH) 128

Hz, C₅(CH₃)₅).

Synthesis of the Mononuclear Fe(II) Alkynyl Complex (η²-

dppe)(η⁵-C₅Me₅)Fe-CtC-2,4,6-(C₆H₂(CH₃)₃) (3b). (η²-dppe)(η⁵-

C₅Me₅)Fe(Cl) (2, 1.000 g, 1.60 mmol), NH₄PF₆(0.315 g, 1.93

mmol), and (2,4,6-trimethylphenyl)acetylene (0.350 g, 2.40 mmol)

were suspended in 30 mL of methanol, and the mixture was stirred

for 48 h at 25 °C. After concentration to 10 mL and decantation,

the pale orange solid that formed was filtrated, washed with

methanol (5 mL), and extracted with 50 mL of dichloromethane.

Concentration of the extract (2 mL) and precipitation by excess

diethyl ether (20 mL) allowed the isolation of [(η²-dppe)(η⁵-C₅-

Me₅)FedCdCH{2,4,6-(C₆H₂)(CH₃)₃}][PF₆] (7b[PF₆]) as an air-

sensitive orange solid (1.256 g, 1.429 mmol, 89%). This vinylidene

complex (1.000 g, 1.13 mmol) was then stirred for 12 h in THF in

the presence of excess potassium tert-butoxide (0.193 g, 1.700

mmol). After evacuation of the solvent and extraction with toluene,

concentration of the extract to dryness, and subsequent washings

with n-pentane, the desired orange complex (η²-dppe)(η⁵-C₅Me₅)-

Fe[CtC-2,4,6-(C₆H₂)(CH₃)₃] (3b) was obtained (0.710 g, 0.965

Complexes 1a-j[PF₆]¹⁵and 4[PF₆]^10awere obtained as previ-

ously reported, and the new complex m-1d[PF₆] was obtained

following a similar route from the known m-1d Fe(II) precursor.²⁸

The organic arylacetylides 5a,b were synthesized following classical

routes either from trimethylsilyl acetylene and meta-xylyl bromide

via a Sonogashira coupling protocol for 5a⁸²or via a Wittig

procedure from mesityl aldehyde and the corresponding dibro-

mocarbene phosphonium salt followed by in situ dehydrohaloge-

nation for 5b.⁸³

(81) Connelly, N. G.; Geiger, W. E. Chem. ReV. 1996, 96, 877-910.

(82) Lavastre, O.; Ollivier, L.; Dixneuf, P. H.; Sinbandhit, S. Tetrahedron

1995, 52, 5495-5504.

(83) (a) Dolhem, F.; Lie`vre, C.; Demailly, G. Tetrahedron Lett. 2002,

43, 1847-1849. (b) Michel, P.; Gennet, D.; Rassat, A. Tetrahedron Lett.

1999, 40, 8575-8578.

894 Organometallics, Vol. 26, No. 4, 2007

Paul et al.

Table 14. Selected Bond Lengths (Å) and Angles (deg) for 1c[PF₆], m-1d[PF₆], 3a, 3b‚C₆H₆, and 3b[PF₆]‚C₆H₆

1c[PF₆]

m-1d[PF₆]

3a

3b‚C₆H₆

3b[PF₆]‚¹/₂CHCl₃

Selected Bond Lengths

1.733

2.1707(13)

2.1788(14)

1.896(5)

1.210(7)

1.455 (7)

1.397(7)

1.392(7)

1.379(7)

1.388(8)

1.400(7)

Fe-(Cp*)_centroid

Fe-P1

1.772

1.776

1.741

1.772

2.2708(7)

2.2565(7)

1.896(3)

1.220(4)

1.437(4)

1.400(4)

1.386(4)

1.390(4)

1.386(4)

1.385(4)

1.404(4)

2.2523(7)

2.2879(7)

1.891(2)

1.207(3)

1.446(3)

1.394(3)

1.383(3)

1.370(4)

1.380(4)

1.381(3)

1.405(3)

2.1740(12)

2.1615(13)

1.899(5)

1.235(6)

1.444(6)

1.408(6)

1.397(6)

1.387(7)

1.393(7)

1.392(6)

1.410(6)

2.2595(11)

2.2441(10)

1.885(4)

1.218(6)

1.436(5)

1.416(6)

1.398(6)

1.395(7)

1.378(7)

1.397(6)

1.411(6)

Fe-P2

Fe-C37

C37-C38

C38-C39

C39-C40

C40-C41

C41-C42

C42-C43

C43-C44

C39-C44

C41-C45

C42-C45

C45-F1

1.385(8)

1.508(8)

1.497(4)

1.335(4)

1.337(4)

1.321(4)

C45-F1

C41-F1

1.365(3)

C43-C46

C40-C45

C42-C46

C44-C47

1.498(9)

1.503(6)

1.504(7)

1.502(6)

1.510(6)

1.514(6)

1.493(6)

Selected Bond Angles

84.79(5)

82.55(14)

83.61(15)

176.9(4)

171.9(5)

-120.3

P1-Fe-P2

84.31(3)

92.68(8)

83.48(8)

171.9(2)

174.8(32)

-87.2

83.78(2)

83.74(1)

90.58(7)

173.1(2)

173.5(2)

-97.5

85.98(5)

85.94(13)

83.22(13)

179.3(4)

177.7(4)

-136.36

83.80(4)

87.47 (11)

86.51(11)

178.6(4)

177.4(4)

37.9

P1-Fe-C37

P2-Fe-C37

Fe-C37-C38

C37-C38-C39

Fe-(Cp*)_centroid/C39-C40 a

^aDihedral angle (Cp* ) pentamethylcyclopentadienyl ligand).

mmol, 85%). X-ray-quality crystals of 3b‚C₆H₆were grown by slow

evaporation of a benzene solution of the complex (see Table 14

for selected bond lengths and angles). Total yield: 76%. Color:

red-orange. MS (LSI⁺, m-NBA): m/z 718.2585 ([M]⁺, 100%), m/z

calc for [C₄₇H₅₀P₂⁵⁶Fe]⁺) 732.2737. Anal. Calcd for C₄₇H₅₀P₂-

Fe₁: C, 77.05; H, 6.88. Found: C, 76.38; H, 6.79. FT-IR (ν, KBr/

CH₂Cl₂, cm^-1): 2036/2037 (s, CtC). Raman (neat, ν, cm^-1): 2037

(vs, CtC). ³¹P{¹H} NMR (C₆D₆, 81 MHz, ppm, δ in ppm): 102.5

(s). ¹H NMR (C₆D₆, 200 MHz, δ in ppm): 7.92 (m, 4H,

[(η²-dppe)(η⁵-C₅Me₅)Fe-CtC-3,5-(C₆H₄(CH₃)₂)][PF₆] (3a-

[PF₆]). Yield: 96%. Color: red-orange. Anal. Calcd for C₄₆H₄₈P₃F₆-

Fe₁: C, 63.97; H, 5.60; F, 13.20. Found: C, 62.90; H, 5.64; F,

13.27. FT-IR (ν, KBr/CH₂Cl₂, cm^-1): 1999 (vw, CtC), 1951 (sh)/

1998 (m, CtC). Raman (neat, ν, cm^-1): 1949 (vw, CtC). ESR

(77 K): g₁) 1.981, g₂) 2.056, g₃) 2.455. UV-vis (CH₂Cl₂):

λ

_max(ꢀ/10³dm³M^-1cm^-1) 252 (45.4); 340 (sh, 6.7); 587 (2.3); 681

(3.8).

[(η²-dppe)(η⁵-C₅Me₅)Fe-CtC-2,4,6-(C₆H₂(CH₃)₃)][PF₆] (3b-

H

_{ortho/Ph1/dppe}); 7.36 (m, 2H, H_Mes); 7.25-7.00 (m, 16H, H_Ar/dppe);

[PF₆]). Yield: 94%. Color: red-brown. Anal. Calcd for C₄₇H₅₀P₃F₆-

Fe₁: C, 64.32; H, 5.74; F, 12.99. Found: C, 64.02; H, 5.66; F,

12.47. IR (KBr/CH₂Cl₂, ν, cm^-1): 1990/1986 (s, CtC), 1957/1954

(w, CtC). Raman (neat, ν, cm^-1): 1990 (s, CtC), 1960 (w, Ct

C). ESR (77 K): g₁) 1.983, g₂) 2.032, g₃) 2.420. UV-vis

(λ_max, CH₂Cl₂, ꢀ/10³dm³M^-1cm^-1): 316 (sh, 20.3); 394 (sh, 6.7);

616 (4.4); 714 (11.8). X-ray-quality crystals of 3b[PF₆] were grown

by slow evaporation of a CHCl₃solution of the compound (see

Table 14 for selected bond lengths and angles).

[(η²-dppe)(η⁵-C₅Me₅)Fe-CtC-3-(C₆H₄)F][PF₆] (m-1d[PF₆]).

Yield: 78%. Color: brown. Anal. Calcd for C₄₄H₄₃F₇P₃Fe₁: C,

61.89; H, 5.08. Found: C, 61.95; H, 5.14. FT-IR (KBr/CH₂Cl₂, ν,

cm^-1): 2012 (w, CtC)/2016 (vw, CtC). ESR (77 K): g₁) 1.975,

g₂) 2.030, g₃) 2.454. X-ray-quality crystals of m-1d[PF₆] were

grown by slow diffusion of n-pentane in a CH₂Cl₂solution of the

complex (see Table 14 for selected bond lengths and angles).

Computational Details. DFT calculations were carried out using

the Amsterdam density functional ADF 2004.01 program⁸⁴on the

model compounds (PH₃)₂(η⁵-C₅H₅)Fe(CtC-4-C₆H₄X)⁺(X ) NO₂,

H, Me, NH₂) and on the real compounds (η²-dppe)(η⁵-C₅Me₅)Fe-

(CtC-4-C₆H₄X)⁺(X ) NO₂, Me). Electron correlation was treated

within the local density approximation (LDA) in the Vosko-Wilk-

Nusair parametrization.⁸⁵The nonlocal corrections of Becke⁸⁶and

2.87 (m, 2H, CH_2dppe); 2.22 (s, 3H, CH₃); 2.13 (s, 6H, 2CH₃); 1.92

(m, 2H, CH_2dppe); 1.57 (s, 15H, C₅(CH₃)₅). ¹³C{¹H} NMR (CDCl₃,

50 MHz, δ in ppm): 141.0-127.5 (m, 8C_Ar/dppe+ 2 CH_Ar/Mes

+

C

_Ar/Mes; + Fe-CtC); 139.2 (s, C_quat./Mes); 131.1 (s, C_quat./Mes); 118.9

(s, Fe-CtC); 87.9 (s, C₅(CH₃)₅); 31.0 (m, CH_2/dppe); 21.6 (s, 2

CH₃); 21.5 (s, CH₃); 10.7 (s, C₅(CH₃)₅). CV: E₀(∆E_p, i_pa/i_pc) -0.22

V (0.08, 1) vs SCE. UV-vis (CH₂Cl₂): λ_max(ꢀ/10³dm³M^-1cm^-1

)

352 (14.2); 496 (0.7).

Selected Data for [(η²-dppe)(η⁵-C₅Me₅)FedCdCH{2,4,6-

(C₆H₂)(CH₃)₃}][PF₆] (7b[PF₆]). FT-IR (ν, KBr, cm^-1): 1632 (s,

FedCdC). ³¹P{¹H } NMR (CDCl₃, 81 MHz, δ in ppm): 92.5 (s,

dppe); 143.1 (septuplet, J_PF) 713 Hz, PF₆^-). H NMR (CDCl₃,

1

200 MHz, δ in ppm): 7.65-7.00 (m, 20H, H_Ar/dppe); 6.77 (m, 2H,

_Mes); 5.01 (t, J_HP) 4.2 Hz, 1H, CdC(Ar)H); 2.93 (m, 2H,

4

H

CH_2dppe); 2.55 (m, 2H, CH_2dppe), 2.28 (s, 3H, CH₃); 1.71 (s, 6H,

CH₃); 1.56 (s, 15H, C₅(CH₃)₅).

General Procedure for the Synthesis of the Mononuclear Fe-

(III) Alkynyl Complexes. A 0.95 equiv. portion of [Fe(η⁵-C₅H₅)₂]-

[PF₆] (0.120 g, 0.361 mmol) was added to a solution of the

corresponding Fe(II) parent (0.380 mmol) in 15 mL of dichlo-

romethane, resulting in an instantaneous darkening of the solution.

Stirring was maintained 1 h at room temperature, and the solution

was concentrated in vacuo to approximatively 5 mL. Addition of

50 mL of n-pentane allowed precipitation of a dark solid. Decanta-

tion and subsequent washings with 3 × 3 mL portions of toluene

followed by 3 × 3 mL of diethyl ether and drying under vacuum

yielded the desired complex [(η²-dppe)(η⁵-C₅Me₅)Fe-CtC-4-

(C₆H₄)X][PF₆] as an analytically pure sample.

(84) (a) te Velde, G.; Bickelhaupt, F. M.; Fonseca, Guerra, C.; van

Gisbergen, S. J. A.; Baerends, E. J.; Snijders, J.; Ziegler, T. Theor. Chem.

Acc. 2001, 22, 931-967. (b) Fonseca, Guerra, C.; Snijders, J.; te Velde,

G.; Baerends, E. J. Theor. Chem. Acc. 1998, 99, 391-403. (c) ADF2.3 and

ADF2002.01; Theoretical Chemistry, Vrije Universiteit: Amsterdam, The

Netherlands, SCM.

Spin Delocalization in Electron-Rich Iron(III) σ-Acetylides

Organometallics, Vol. 26, No. 4, 2007 895

Table 15. Crystal Data, Data Collection, and Refinement Parameters for 1c[PF₆], m-1d[PF₆], 3a, 3b‚C₆H₆, and

3b[PF₆]‚1/2CHCl₃

1c[PF₆]

m-1d[PF₆]

3a

3b‚C₆H₆

3b[PF₆]‚1/2CHCl₃

formula

fw

C₄₅H₄₃F₉P₃Fe₁

903.55

C₄₄H₄₃F₇P₃Fe₁

853.54

C₄₆H₄₈P₂Fe₁

718.63

C₄₇H₅₀P₂Fe₁‚C₆H₆

810.77

C₄₇H₅₀P₂Fe₁,PF₆‚1/2CHCl₃

937.31

temp (K)

cryst syst

space group

a (Å)

b (Å)

c (Å)

120(1)

orthorhombic

Pbcn

100(1)

monoclinic

P2₁

120(1)

triclinic

P1h

120(1)

triclinic

P1h

120(1)

triclinic

P1h

16.0682(5)

12.2876(4)

42.3400(10)

90.00(0)

8359.6(2)

8

9.5078(4)

16.2192(6)

13.2545(6)

90.00(0)

106.059(3)

90.00(0)

1964.2(1)

2

8.5747(5)

11.7050(7)

19.2720(10)

88.957(3)

83.951(3)

79.234(2)

1889.6(2)

2

10.6961(4)

11.0582(3)

20.3047(8)

80.010(2)

86.082(2)

64.282(2)

2130.8(2)

2

11.0094(3)

15.0407(3)

15.8834(3)

67.975(1)

79.233(1)

74.729(1)

2340.7(1)

2

R (deg)

â (deg)

γ (deg)

V (Å³)

Z

D

_calcd(g cm^-3

)

1.436

1.443

1.263

1.264

1.330

cryst size (mm)

F(000)

0.35 × 0.25 × 0.22

0.22 × 0.12 × 0.10

0.22 × 0.12 × 0.12

0.22 × 0.22 × 0.18

0.45 × 0.32 × 0.32

3720

882

760

860

972

diffractometer

radiation

Oxford Diffraction Xcalibur Saphir 3

KappaCCD (Nonius)

Mo KR

KappaCCD (Nonius)

Mo KR

KappaCCD (Nonius)

Mo KR

0.548

54

Mo KR

0.572

54

abs coeff (mm^-1

data collection:

)

0.515

0.465

0.566

60

54

θ

_max(deg)

frames

1145

210

Ω rotation (deg)

seconds/frame

θ range

0.7

20

0.7

20

0.3

12

1.06-27.65

0/11

-14/15

-24/24

16 215

7942

1.5

30

2.0

10

1.39-27.50

0/14

-18/19

-19/20

42 484

10 730

8193

2.71-32.22

-23/23

-10/18

-60/62

75 304

13 873

8428

2.98-32.26

-13/14

-16/23

-19/19

19 221

9134

1.02-27.56

-13/13

-14/14

-26/67

28 978

9705

h k l range

no. total reflns

no. unique reflns

no. obsd reflns

[I > 2σ(I)]

6764

5867

6589

no. restraints/params

0/526

a ) 0.0745

0/496

a ) 0.0370

0/443

a ) 0.0904

0/560

a ) 0.057

0/550

a ) 0.1451

w ) 1/[σ²(F_o)²+

(aP)²+ bP]

2

(where P ) [F_o

+

b ) 6.9722

b ) 0.0000

b ) 5.4105

b ) 6.66

b ) 6.0134

2

F_c]/3)

final R

R_w

0.065

0.155

0.109

0.165

1.050

1.711

-1.251

0.033

0.068

0.053

0.071

0.892

0.402

-0.385

0.087

0.206

0.123

0.236

1.139

0.619

-0.818

0.072

0.180

0.113

0.201

1.082

0.539

-0.568

0.081

0.225

0.105

0.251

1.057

2.314

-1.015

R indices (all data)

_w(all data)

goodness of fit/F²(S_w)

∆F_max(e Å^-3

∆F_min(e Å^-3

R

)

of Perdew⁸⁷were added to the exchange and correlation energies,

respectively. The numerical integration procedure applied for the

calculations was developed by te Velde et al.⁸⁵The basis set used

for the metal atom was a triple-ú Slater-type orbital (STO) basis

for Fe 3d and 4s and a single-ú function for Fe 4p. A triple-ú STO

basis set was employed for H 1s and for 2s and 2p of C, N, and O,

extended with a single-ú polarization function (2p for H; 3d for C,

N, and O). Full geometry optimizations (assuming C₁symmetry)

were carried out on each complex, using the analytical gradient

method implemented by Verluis and Ziegler.⁸⁸For compound 1f-

H⁺(X ) H), no local minimum was found for the perpendicular

conformation and the C_ssymmetry was imposed. The energy

difference between these conformers is therefore meaningless and

was not reported. Spin-unrestricted calculations were performed

for all the considered open-shell systems. Representations of the

molecular orbitals were done using MOLEKEL4.1.⁸⁹

or on a NONIUS Kappa CCD with graphite-monochromatized Mo

KR radiation. The cell parameters were obtained with Denzo and

Scalepack with 10 frames (psi rotation: 1° per frames).⁹⁰The data

collection⁹¹(2θ_max, number of frames, Ω rotation, scan rate, and

HKL range as given in Table 14) provided reflections for 1c[PF₆],

m-1d[PF₆], 3a, 3b‚C₆H₆, and 3b[PF₆]‚¹/₂CHCl₃. Subsequent data

reduction with Denzo and Scalepack⁹⁰gave the independent

reflections (Table 15). The structures were solved with SIR-97,

which revealed the non-hydrogen atoms.⁹²After anisotropic refine-

ment, the remaining atoms were found in Fourrier difference maps.

The complete structures were then refined with SHELXL97⁹³by

the full-matrix least-squares technique (use of F²magnitude; x, y,

z, â_ijfor Fe, P, C, N, and/or O atoms, x, y, z in riding mode for H

atoms with variables “N(var.)”, observations and “w” used as

(89) Flu¨kiger, P.; Lu¨thi, H. P.; Portmann, S.; Weber, J. MOLEKEL4.1;

Swiss Center for Scientific Computing (CSCS): Switzerland, 2000-

2001.

(90) Otwinowski, Z.; Minor, W. In Methods in Enzymology; Carter, C.

W., Sweet, R. M., Eds.; Academic Press: London, 1997; Vol. 276, pp 307-

326.

Crystallography. Crystals of 1c[PF₆], m-1d[PF₆], 3a, 3b‚C₆H₆,

and 3b[PF₆]‚¹/₂CHCl₃were obtained as described above. The

samples were studied on an Oxford Diffraction Xcalibur Saphir 3

(85) Vosko, S. D.; Wilk, L.; Nusair, M. Can. J. Chem. 1990, 58, 1200-

1211.

(91) Nonius, B. V. Kappa CCD Software; Delft: The Netherlands,

1999.

(92) Altomare, A.; Burla, M. C.; Camalli, M.; Cascarano, G.; Giacovazzo,

C.; Guagliardi, A.; Moliterni, A. G. G.; Polidori, G.; Spagna, R. J. Appl.

Chem. 1998, 31, 74-77.

(93) Sheldrick, G. M. SHELX97-2, Program for the refinement of crystal

structures; Univ. of Go¨ttingen: Germany, 1997.

(86) (a) Becke, A. D. J. Chem. Phys. 1986, 84, 4524-4529. (b) Becke,

A. D. Phys. ReV. A 1988, 38, 3098-3100.

(87) (a) Perdew, J. P. Phys. ReV. B 1986, 33, 8822-8824. (b) Perdew,

J. P. Phys. ReV. B 1986, 34, 7406.

(88) Versluis, L.; Ziegler, T. J. Chem. Phys. 1988, 88, 322-328.

896 Organometallics, Vol. 26, No. 4, 2007

Paul et al.

defined in Table 15). Atomic scattering factors were taken from

the literature.⁹⁴ORTEP views of 1c⁺, m-1d⁺, 3a, and 3b⁺were

realized with PLATON98.⁹⁵

Supporting Information Available: Experimental details about

the NMR measurements (pulse widths, etc.), H NMR shifts vs

1

Fe(II)/Fe(III) ratios in CD₂Cl₂at 25 °C for 1g/1g[PF₆] and 2/2[PF₆],

additional details regarding the assignment of the ¹H and ¹³C NMR

shifts, COSY, HMQC, and NOESY spectra for selected compounds,

T₁measurements for 1a[PF₆] and 1b[PF₆], half-width data for

selected compounds, Solomon-Bloembergen treatment of ¹H NMR

relaxation, selected correlations, and CIF files for 1c[PF₆], m-1d[PF₆],

3a, 3b‚C₆H₆, and 3b[PF₆]‚CDCl₃. This material is available free

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

Acknowledgment. F.P. and N.G. thank Region Bretagne for

financial support. The CNRS is also acknowledged for financial

support and the Institut de De´veloppement et de Ressources en

Informatique (IDRIS, Orsay) for computing facilities.

(94) Reidel, D. International Tables for X-ray Crystallography; Kynoch

Press (present distrib. D. Reidel, Dordrecht): Birmingham, 1974; Vol. IV.

(95) Spek, A. L. PLATON, A Multipurpose Crystallographic Tool;

Utrecht University: Utrecht, The Netherlands, 1998.

OM060989J

Article Doi

Spin delocalization in electron-rich iron(III) piano-stool σ-acetylides. An experimental (NMR) and theoretical (DFT) investigation

DOI: 10.1021/om060989j

Source and publish data:

Authors:

Article abstract of DOI:10.1021/om060989j

Full text of DOI:10.1021/om060989j

Products guided by the article

R&D Labs maybe for 928264-70-4

Relevant to this article

Hot Product