Electron-transfer chain catalysis in phosphine replacement reaction: Determination of relative donor capability of arylpyridylphosphines

Article abstract of DOI:10.1016/j.ica.2006.12.051

The slight differences in the donor capabilities of PPh_nPy_3-n (n = 0-3) could be measured directly in the equilibrium of phosphine replacement reaction on (η⁵-C₅H₅)Fe(CO)₂-phosphi ne complexes, taking advantage of the radical pathway to establish equilibrium rapidly. The simultaneous determination of equilibrium constants is done in a single experiment. The donor capability increases in the order PPh₃ < PPh₂Py < PPhPy₂ < PPy₃ with quantified (η⁵ - C₅ H₅) Fe (CO)₂⁺-affinity scales at 1, 4.90, 11.0, and 20.3, respectively.

Full text of DOI:10.1016/j.ica.2006.12.051

Inorganica Chimica Acta 360 (2007) 2464–2470
www.elsevier.com/locate/ica
Electron-transfer chain catalysis in phosphine replacement reaction:
Determination of relative donor capability of arylpyridylphosphines
a,b,
a,c
Olalere G. Adeyemi
*, Ling-Kang Liu
a
Institute of Chemistry, Academia Sinica, Nankang, Taipei 11529, Taiwan, ROC
Department of Chemical Sciences, College of Natural Sciences, Redeemer’s University, Km 46, Lagos/Ibadan Expressway, Redemption City, Nigeria
b
c
Department of Chemistry, National Taiwan University, Taipei 10767, Taiwan, ROC
Received 16 August 2006; received in revised form 4 December 2006; accepted 8 December 2006
Available online 20 January 2007
Abstract
The slight differences in the donor capabilities of PPhnPy3ꢀn (n = 0–3) could be measured directly in the equilibrium of phosphine
replacement reaction on (g5-C5H5)Fe(CO)2-phosphine complexes, taking advantage of the radical pathway to establish equilibrium rap-
idly. The simultaneous determination of equilibrium constants is done in a single experiment. The donor capability increases in the order
PPh3 < PPh2Py < PPhPy2 < PPy3 with quantified ðg5-C5H5ÞFeðCOÞ2þ-affinity scales at 1, 4.90, 11.0, and 20.3, respectively.
ꢀ 2006 Elsevier B.V. All rights reserved.
Keywords: Electron-transfer chain catalysis; Arylpyridylphosphines; Iron complexes; Radical process; Phosphine exchange; Cyclic voltammetry
1. Introduction
oretical [10], and electrochemical [11] methods and X-ray
crystallography [12].
Organophosphines are used in many metal mediated
reactions [1]. As a consequence, chemists have been inter-
ested in deducing their stereo-electronic properties [2],
among which the best known parameters are the cone angle
H and the electronic parameter v by Tolmþan [3]. Also a
good parameter, the pKa value for HPR3 depends on
the interaction between the phosphorus and H+, a hard
acid [4]. Such an interaction is generally considered not
the same as that between the phosphorus ligand and a
low valent metal center in an organometallic complex.
Derived from the IR data for LNi(CO)3 (L = phosphorus
ligand), the v values reflect a combined property of both
r-donor and p-acceptor [3]. Many other evaluations have
been attempted using IR [5], NMR [6], UV/Vis [7], Photo-
electron [8], and Mo¨ssbauer spectroscopy [9], as well as the-
Giering et al. reported in their quantitative analysis of
ligand effects, a series of phosphines with H, pKa, v, mCO
and Eꢁ values were tabulated. In particular, the mCO and
Eꢁ values were obtained from the iron-complexes (g5-
C5H5)Fe(CO)(L)C(O)Me and (g5–MeC5H4)Fe(CO)(L)-
C(O)Me, where L is a phosphorus ligand [13]. For the
following triarylphosphines with H = 145ꢁ, e.g., P(p-
C6H4NMe2)3, P(p-C6H4OPh)3, P(p-C6H4Me)3, PPh3, P(p-
C6H4F)3, P(p-C6H4Cl)3, P(p-C6H4CF3)3, a decrease of
ligand donor capability was clearly evident and there was
a nice correlation among the parameters.
The series of arylpyridylphosphines PPhnPy3ꢀn (n = 0–3)
are also with H values approximately 145ꢁ, yet without read-
ily available pKa, v, mCO and Eꢁ values for the evaluation of
their donor capability. The only starting point is that the
31P NMR chemical shifts are more upfield with more Ph-
groups (d ꢀ0.6, ꢀ1.9, ꢀ3.2, and ꢀ4.7 with increasing num-
ber of Ph-groups), a fact not necessarily translated into
increasing donor capability, however. As there are slight dif-
ferences in the donor capabilities of PPhnPy3ꢀn, we have
found that the direct measurement of an equilibrium
*
Corresponding author. Address: Department of Chemical Sciences,
College of Natural Sciences, Redeemer’s University, Km 46, Lagos/
Ibadan Expressway, Redemption City, Nigeria. Tel.: +234 805 5516450.
E-mail addresses: adeyemio@run.edu.ng, drlereadeyemi@yahoo.com
(O.G. Adeyemi).
0020-1693/$ - see front matter ꢀ 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.ica.2006.12.051

O.G. Adeyemi, L.-K. Liu / Inorganica Chimica Acta 360 (2007) 2464–2470
2465
constant in the phosphine replacement reaction on (g5-
C5H5)Fe(CO)2-phosphinecomplexispossible,takingadvan-
tage of the radical pathway to establish equilibrium rapidly.
A simultaneous determination of equilibrium constants is
also likely so that their donor capabilities could be compared
in a single experiment.
With easily available ½ðg5-C5H5ÞFeðCOÞ2ðPPh3Þþꢂ
½PF6ꢀꢂ, the 17e–19e pathway enables one to instantly dif-
ferentiate an incoming PPhnPy3ꢀn replacement for PPh3
on ½ðg5-C5H5ÞFeðCOÞ2PPh3þꢂ. Scheme 1 shows such an
ETC reaction, the equilibrium constant K of Eq. (1) being
the product of Kredox of Eq. (2) and Kexchange of Eq. (3).
The former is the coupling between the reduction of
(g5-C5H5)Fe(CO)2P1+ cation to (g5-C5H5)Fe(CO)2P1Å
radical and the oxidation of (g5-C5H5)Fe(CO)2P2Å radical
to (g5-C5H5)Fe(CO)2P2+ cation. The latter is the coupling
between an association of P2 with ðg5-C5H5ÞFeðCOÞ2Å
radical to form (g5-C5H5)Fe(CO)2P2Å radical and a disso-
ciation of (g5-C5H5)Fe(CO)2P1Å radical to form P1 and
ðg5-C5H5ÞFeðCOÞÅ2 radical. Both redox and exchange are
facile because of high energy radical process in nature
[15].
The following is a qualitative analysis on the Kredox. If P2
is a better donor than P1, the Fe atom of (g5-C5H5)Fe-
(CO)2P2Å radical has a higher electron density than that of
(g5-C5H5)Fe(CO)2P1Å radical, i.e., the (g5-C5H5)Fe-
(CO)2P2Å radical is likely to be a better reductant than the
(g5-C5H5)Fe(CO)2P1Å radical and one would expect that
the Kredox for Eq. (2) must be greater than 1. On the other
hand, Kexchange could also be analyzed in a qualitative way
with Eq. (3) envisioned as P1 and P2 in competition for the
1.1. Radical process
Pioneered by Rich and Taube in inorganic chemistry
[14], the electron-transfer chain catalysis (ETC) is applied
to the preparation of organometallic complexes, utilizing
the greatly enhanced reactivity of organometallic 17e and
19e radicals in comparison to their 16e and 18e analogues
[15–17]. The PPh3 substitution for CO on ½ðg5-C5H5Þ
FeðCOÞ3þꢂ proceeds with 17e–19e pathway after electro-
chemical or chemical initiation, for instance, and yields
½ðg5-C5H5ÞFeðCOÞ2PR3þꢂ quantitatively [18]. The reaction
of 1:1 (g5-C5H5)Fe(CO)2I/PPh3 in THF at ꢀ78 ꢁC, initial-
ized by the addition of a small amount of reducing agents,
e.g., diluted RLi [19] or (g5-C5H5)2Co, also follows the
17e–19e pathway and gives quantitative precipitates of
[(g5-C5H5)Fe(CO)2(PPh3)+][Iꢀ] instantaneously, Iꢀ being
selectively replaced by PPh3 [20]. The anion [Iꢀ] could be
exchanged into ½PF6ꢀꢂ without difficulty.
[(η5-C5H5)Fe(CO)2P2+][P1]
[(η5-C5H5)Fe(CO)2P1+][P2]
P2
Fe +
Fe +
(Eq. 1)
K
=
P1
OC
OC
P2
P1
OC
OC
[(η5-C5H5)Fe(CO)2P1+] + [(η5-C5H5)Fe(CO)2P2.]
[(η5-C5H5)Fe(CO)2P1.] + [(η5-C5H5)Fe(CO)2P2+]
Fe+
Fe +
OC
OC
P1
P2
OC
K redox
=
OC
+e-
[(η5-C5H5)Fe(CO)2P1.][(η5-C5H5)Fe(CO)2P2+]
[(η5-C5H5)Fe(CO)2P1+][(η5-C5H5)Fe(CO)2P2.]
(Eq. 2)
.
.
Fe
Fe
OC
OC
P2
19e-
P1
[(η5-C5H5)Fe(CO)2P1.] + P2
P1 + [(η5-C5H5)Fe(CO)2P2.]
OC
OC
19e-
K exchange
[P1][( 5-C H )Fe(CO) P2.]
η
5
5
2
(Eq. 3)
=
[(η5-C5H5)Fe(CO)2P1.][P2]
P1
P2
Scheme 1. Phosphine exchange reaction with an electron-transfer chain catalytic pathway.

2466
O.G. Adeyemi, L.-K. Liu / Inorganica Chimica Acta 360 (2007) 2464–2470
same ½ðg5-C5H5ÞFeðCOÞÅ2 radical. If P2 is a better donor
than P1, the ðg5-C5H5ÞFeðCOÞÅ2 radical would prefer a liga-
tion with P2 to a ligation with P1 and one would also expect
that the Kexchange of Eq. (3) must be greater than 1. These
overall effects of a favoring phosphine substitution on the
thermodynamic side combined with a rapid radical process
on the dynamic side lead us to study the small difference
among ½ðg5-C5H5ÞFeðCOÞ PPhn Py þꢂ by a simple 31P
2.2.1. [(g5-C5H5)Fe(CO)2PPh2Pyþ][PF 6
]
ꢀ
Yield: 1.24 g (78%); mp: 222–223 ꢁC; IR (CH2Cl2): mCO
2062 (s), 2021(s) cmꢀ1
;
31P NMR (acetone-d6): d 70.8 (s),
1
ꢀ143.1 (hep, JPF = 2829 Hz); 1H NMR (acetone-d6): d
8.94 (b, 1H, Py), 7.46–7.96 (m, 13H, Ph and Py), 5.57 (b,
5H, Cp); 13C NMR (acetone-d6):
d
210.39 (d,
2JPC = 25.67 Hz, CO), 126.94–152.07 (m, Ph and Py),
89.28 (s, Cp); MS (m/z): 440 (M+ꢀPF6). Anal. Calc. for
C24H19NFeO2F6P2: C, 49.23; H, 3.25; N, 2.39. Found: C,
49.09; H, 3.14; N, 2.05%.
2
3ꢀn
NMR technique, namely, the intensity integration ratio.
2. Experimental
2.2.2. [(g5-C5H5)Fe(CO)2PPhPy2þ][PF 6
]
ꢀ
2.1. General
Yield: 0.94 g (90%); mp: 205–206 ꢁC; IR (CH2Cl2): mCO
2063 (s), 2021 (s) cmꢀ1
;
31P NMR (acetone-d6): d 74.2 (s),
1
All manipulations were performed under an atmosphere
of prepurified nitrogen with standard Schlenk techniques.
All solvents were distilled from an appropriate drying agent
[21]. Infrared spectra were recorded in CH2Cl2 using CaF2
ꢀ143.1 (hep, JPF = 2829 Hz); 1H NMR (acetone-d6): d
3
8.96 (d, 2H, JPH = 4.04 Hz, Py), 7.42–7.99 (m, 11H, Ph
and Py), 5.58 (b, 5H, Cp); 13C NMR (acetone-d6): d
2
210.38 (d, JPC = 24.45 Hz, CO), 127.00–155.98 (m, Ph
1
optics on a Perkin–Elmer 852 spectrophotometer. The H
and Py), 89.03 (s, Cp); MS (m/z): 442 (M+ꢀPF6). Anal.
Calc. for C23H18N2FeO2F6P2: C 47.10; H, 3.07; N, 4.78.
Found: C, 46.33; H, 2.84; N, 4.36%.
NMR and 13C NMR spectra were obtained on Bruker
AC200/AC300 spectrometers, with chemical shifts reported
in d values, downfield positive, relative to the residual sol-
vent resonance of CDCl3 (1H d 7.24, 13C d 77.0). The 31P
NMR spectra were obtained on Bruker AC200/AC300
spectrophotometer using 85% H3PO4 as an external stan-
dard (d 0.00). The melting points were determined on a
Yanaco MPL melting-point apparatus and uncorrected.
(g5-C5H5)Fe(CO)2I [22], (g5-C5H5)Fe(CO)2Me [23],
PPh2Py [24], PPhPy2 [25], PPy3 [25], ½ðg5-C5H5ÞFeðCOÞ2-
PPh3þꢂ½PF6ꢀꢂ [20,26] and (g5-C5H5)Fe(CO)C(O)Me(PPh3)
[27] were prepared according to the literature procedure.
Other reagents were obtained from commercial sources
and used without further purification.
ꢀ
2.2.3. [(g5-C5H5)Fe(CO)2PPy3þ][PF 6
]
Yield: 0.45 g (77%); mp: 200–202 ꢁC; IR (CH2Cl2): mCO
2061 (s), 2020 (s) cmꢀ1
;
31P NMR (acetone-d6): d 76.1 (s),
1
ꢀ143.1 (hep, JPF = 2829 Hz), 1H NMR (acetone-d6): d
8.95 (b, 3H, Py) 7.51–7.98 (m, 9H, Py), 5.55 (b, 5H, Cp);
13C NMR (acetone-d6): d 210.22 (d, JPC = 24.00 Hz,
2
CO), 127.00–155.75 (m, Py), 88.67 (s, Cp); MS (m/z): 443
(M+ꢀPF6+1). Anal. Calc. for C22H17N3FeO2F6P2: C
44.97; H, 2.90; N, 7.16. Found: C, 44.73; H, 2.72; N, 6.84%.
2.3. Phosphine exchange under ETC conditions
2.2. Preparation of [(g5-C5H5)Fe(CO)2Lþ][PF 6ꢀ],
L = PPh2Py, PPhPy2, PPy3
½ðg5-C5H5ÞFeðCOÞ2PPhnPy3ꢀnþꢂ½PF6ꢀꢂ (0.45 mmol) and
PPhmPy3ꢀm (0.15 mmol · 3, m = 0–3, m ¼ n) were com-
pletely dissolved in CH3CN (1.5 mL) and transferred to a
10 mm co-axial NMR tube that has an inner tube filled
with 0.30 mmol PPh3(O) in CDCl3 (0.5 mL) for external
reference (d = 29.5 in 31P NMR spectra) and for use in
locking of the magnetic field. The 3:1:1:1 intensity integra-
tion ratio for the cation and the free ligands was observed
from the 31P NMR spectra. A trace amount of (g5-
C5H5)2Co was then introduced to the solution in the
NMR tube. The 31P NMR spectrum was remeasured in
The procedures are the same for all PPhnPy3ꢀn, follow-
ing the preparation of ½ðg5-C5H5ÞFeðCOÞ2PPh3þꢂ½PF6ꢀꢂ in
the literature [20]. Only L = PPh2Py is shown here as a rou-
tine run. (g5-C5H5)Fe(CO)2I (0.86 g, 2.83 mmol) and
PPh2Py (0.74 g, 2.83 mmol) were completely dissolved in
THF (30 mL) and maintained at ꢀ78ꢁ C. A yellow precip-
itate appeared on addition of a trace amount of n-BuLi
(1.6 M in hexane, 2 drops) to the solution. The solution
was further stirred for 15 min before being gradually
warmed up to room temperature. After filtration and wash-
ing with Et2O for several times, the precipitate together
with excess NH4PF6 was redissolved in CH3CN with vigor-
ous stirring. Addition of sufficient H2O gave precipitates
which were filtered and then washed with H2O and Et2O
to produce ½ðg5-C5H5ÞFeðCOÞ PPh2Pyþꢂ½PF6ꢀꢂ. For
½ðg5-C5H5ÞFeðCOÞ PPhPy þꢂ½PF6ꢀ2ꢂ, (g5-C5H5)Fe(CO)2I
which
a
total of eight peaks corresponding to
½ðg5-C5H5ÞFeðCOÞ2PPhmPy3ꢀmþꢂ and PPhnPy3ꢀn (m,
n = 0–3) appeared. The intensity integration ratio was
recorded again. The results are shown in Table 1.
2.4. Preparation of (g5-C5H5)Fe(CO)C(O)Me(L),
L = PPh3, PPh2Py, PPhPy2, PPy3
2
2
(0.54 g, 1.78 mmol), PPhPy2 (0.47 g, 1.78 mmol), THF
(26 mL) and for ½ðg5-C5H5ÞFeðCOÞ2PPy3þꢂ½PF6ꢀꢂ, (g5-
C5H5)Fe(CO)2I (0.313 g, 1.0 mmol), PPy3 (0.266 g,
1.0 mmol), THF (12 mL).
(g5-C5H5)Fe(CO)2Me (0.50 g, 2.5 mmol) and the ligand
L (2.5 mmol) were dissolved in THF (10 mL, freshly dis-
tilled) and refluxed for 48 h. The mixture, an orange solu-
tion, was allowed to cool down to room temperature

O.G. Adeyemi, L.-K. Liu / Inorganica Chimica Acta 360 (2007) 2464–2470
2467
Table 1
The measured equilibrium constants of [FpP1]+ + P2 ! P1 + [FpP2]+a
Reactant
PPh3
PPh2Py
PPhPy2
PPy3
þ
FpPPh3
1
1
1
1
4.76b
10.7
11.4
11.2
10.7
18.7
25.6
20.4
16.7
PPh3
5.41c
5.05d
4.36e
K12 = 4.90(44)f
K11 = 1
K13 = 11.0(4)
K14 = 20.3(38)
FpPPh2Py+
0.210
0.185
0.198
0.229
1
1
1
1
2.25
2.11
2.22
2.45
3.93
4.74
4.03
3.69
PPh2Py
PPhPy2
PPy3
K21 = 0.205(19)
K22 = 1
K23 = 2.25(14)
K24 = 4.10(45)
þ
FpPPhPy2
0.0932
0.0876
0.0893
0.0938
0.444
0.474
0.450
0.408
1
1
1
1
1.74
2.24
1.82
1.51
K31 = 0.0910(30)
K32 = 0.444(27)
K33 = 1
K34 = 1.83(31)
þ
FpPPy3
0.0535
0.254
0.575
1
0.0390
0.211
0.446
1
0.0491
0.248
0.549
1
0.0599
0.271
0.662
1
K14 = 0.0504(88)
K42 = 0.246(25)
K43 = 0.558(89)
K44 = 1
FpPPh2Py+
Product
þ
þ
FpPPhPy2
þ
FpPPy3
FpPPh3
a
Fp = (g5-C5H5)Fe(CO)2, P1, P2 = PPh3, PPh2Py, PPhPy2, PPy3.
b
c
d
e
f
With the initial condition: 3FpPPh3þPF6ꢀ þ PPh2Py þ PPhPy2 þ PPy3.
With the initial condition: 3FpPPh2PyþPF6ꢀ þ PPh3 þ PPhPy2 þ PPy3.
With the initial condition: 3FpPPhPy2þPF6ꢀ þ PPh3 þ PPh2Py þ PPy3.
With the initial condition: 3FpPPy3þPF6ꢀ þ PPh3 þ PPh2Py þ PPhPy2.
Average (standard deviation) of experiments b–e.
then the insoluble solid was removed by passing through a
pad of celite. The solvent was removed under vacuum. The
residue was dissolved in minimal CH2Cl2, mixed well with
a small quantity of alumina and packed on the top of an
alumina column. The bands were eluted first with 1:10
EtOAc/hexane to remove the trace amount of yellow (g5-
C5H5)Fe(CO)2Me. The acetyl complexes (g5-C5H5)Fe-
(CO)C(O)Me(L) were collected using eluents of higher
polarity: with 1:5 EtOAc/hexane for L = PPh3; with 1:2
EtOAc/hexane for L = PPh2Py; with 3:1 EtOAc/hexane
for L = PPhPy2; and with 3.5:1 EtOAc/hexane for
L = PPy3.
d 8.66 (b, 2H, Py), 7.20–7.62 (m, 12H, Ph and Py), 4.48
(s, 5H, Cp), 2.41 (s, 3H, Me); 13C NMR (CDCl3): d
276.01 (d, 3JPC = 22 Hz, C(O)Me), 219.92 (d,
3JPC = 30 Hz, CO), 123.14–161.00 (m, Ph and Py), 84.34
(s, Cp), 51.90 (s, Me); MS (m/z): 457 (M++1); Anal. Calc.
for C24H21N2PFeO2: C, 63.16; H, 4.61; N, 6.14. Found: C,
63.51; H, 4.70; N, 5.99%.
(g5-C5H5)Fe(CO)C(O)Me(PPy3): yield 0.58 g (51%);
mp: 127–128 ꢁC; IR (CH2Cl2): mCO 1927 (s) 1597 (m)
1
cmꢀ1
;
31P NMR (CDCl3): d 89.7 (s); H NMR (CDCl3):
d 8.64 (b, 3H, Py), 7.84 (b, 3H, Py), 7.66 (b, 3H, Py),
7.22 (b, 3H, Py), 4.51 (s, 5H, Cp), 2.43 (s, 3H, Me); 13C
3
(g5-C5H5)Fe(CO)C(O)Me(PPh3): yield 0.59 g (59%); IR
NMR (CDCl3): d 275.93 (d, JPC = 23 Hz, C(O)Me),
(CH2Cl2): mCO 1916 (s) 1601 (m) cmꢀ1
;
31P NMR (CDCl3):
219.70 (d, JPC = 30 Hz, CO), 122.76–161.81 (m, Ph and
3
3
d 75.23 (s); Literature [27] IR (CHCl3): mCO 1920 (vs, br),
Py), 84.62 (s, Cp), 51.81 (d, Me, JPC = 4 Hz); MS (m/z):
1598 (s) cmꢀ1
.
458 (M++1); Anal. Calc. for C23H20N3PFeO2: C, 60.39;
(g5-C5H5)Fe(CO)C(O)Me(PPh2Py): yield 0.53 g (47%);
H, 4.38; N, 9.19. Found: C, 60.23, H4.41; N, 9.36%.
mp: 132–133 ꢁC; IR (CH2Cl2): mCO 1921 (s), 1599 (m)
1
cmꢀ1
;
31P NMR (CDCl3): d 80.9 (s); H NMR (CDCl3):
2.5. Cyclic voltammetry study
4
d 8.68 (d, JPH = 3.94 Hz, 1H, Py), 7.18–7.60 (m, 13H,
Ph and Py), 4.44 (s, 5H Cp), 2.38(s, 3H, Me); 13C NMR
(CDCl3): d 276.83 (b, C(O)Me), 220.07 (b, CO), 123.13–
161.14 (m, Ph and Py), 84.96 (s, Cp), 52.18 (s, Me); MS
(m/z): 466 (M++1). Anal. Calc. for C25H22NO2FeP: C,
65.93; H, 4.84; N, 3.08. Found: C, 65.83; H, 4.94; N, 2.87%.
(g5-C5H5)Fe(CO)C(O)Me(PPhPy2): yield 0.55 g (48%);
The cyclic voltammograms of 10ꢀ3 M ½ðg5-C5H5ÞFe-
ðCOÞ2Lþꢂ½PF6ꢀꢂ and (g5-C5H5)Fe(CO)C(O)Me(L) where
L = PPh3, PPh2Py, PPhPy2, and PPy3 were measured at
room temperature on 10ꢀ1 M ½Bu4Nþꢂ½PF ꢀꢂ in CH2Cl2
6
solution. The working electrode was a glassy cabon elec-
trode. The reference electrode was Ag/AgNO3 for use in
non-aqueous systems. The scan rate was 100 mV/s in
the range of ꢀ1500 to 0 mV for [(g5-C5H5)Fe(CO)2L+]
mp: 136–137 ꢁC; IR (CH2Cl2): mCO 1923 (s), 1598 (m)
1
cmꢀ1
;
31P NMR (CDCl3): d 84.6 (s); H NMR (CDCl3):

2468
O.G. Adeyemi, L.-K. Liu / Inorganica Chimica Acta 360 (2007) 2464–2470
complexes and in the range of ꢀ500 to +500 mV/s for ace-
tyl complexes, potentials being relative to ferrocene0,+
(0.281 V). For acetyl complexes with the phosphines bear-
ing at least one Py group, only oxidative waves were
recorded, no corresponding reductive peaks being
observed. For [(g5-C5H5)Fe(CO)2L+] complexes, only
reductive and no oxidative waves were observed. The fol-
lowing is the listing of measured potentials:
The 31P resonances of free PPhnPy3ꢀn ligands are seen to
be more downfield with more Py groups (d ꢀ4.7, ꢀ3.2,
ꢀ1.9, and ꢀ0.6 for 0, 1, 2 and 3 Py-groups, respectively).
Given a Ph-group contribution as rPh and Py-group contri-
bution as rPy, the relationship between the chemical shifts
and the group contributions is linear, namely,
X
dðfree ligandÞ ¼ ꢀ4:65d þ
ri
i
where ri is rPh (=0.00d) and rpy (=1.36d). Upon complex-
ation, the 31P resonances of the acetyl complexes (g5-
C5H5)Fe(CO)C(O)MePPhnPy3ꢀn are d 75.2, 80.9, 84.6,
and 89.7, respectively, in the order of increasing number
of Py-groups. The complexation shifts D (=d of com-
plex ꢀ d of ligand) are 79.9–90.3 d downfield. Employing
the same group contribution ri as for the free ligands, the
31P chemical shifts of the acetyl complexes could be calcu-
lated within 0.7 d, in the linear fitting as follows:
Complex
(g5-C5H5)2Fe0,+
Potential (V)
0.281
0.244
(g5-C5H5)Fe(CO)C(O)Me(PPh3)0,+
(g5-C5H5)Fe(CO)C(O)Me(PPh2Py)0,+
(g5-C5H5)Fe(CO)C(O)Me(PPhPy2)0,+
(g5-C5H5)Fe(CO)C(O)Me(PPy3)0,+
0.227a
0.200a
0.195a
ꢀ1.234b
ꢀ1.353b
ꢀ1.380b
ꢀ1.414b
þ;0
ðg5-C5H5ÞFeðCOÞ2PPh3
(g5-C5H5)Fe(CO)2PPh2Py+,0
!
þ;0
ðg5-C5H5ÞFeðCOÞ PPhPy2
X
ðg5-C5H5ÞFeðCOÞ22PPy3
d
¼ 75:52d þ 3:47
ri
þ;0
ðacetyl complexÞ
i
a
Only oxidative wave observed.
Only reductive wave observed.
b
where a ratio of 3.47 is necessary. Greater than 1.00, this
constant suggests an amplification about the phosphine
group contributions upon complexation. The similar anal-
ysis, when applied to the ½ðg5-C5H5ÞFeðCOÞ2PPhnPy3ꢀnþꢂ
series, results in complexation shifts D ranging from 67.2
to 76.6 d downfield, and a linear fitting of
3. Results and discussion
The preparation of (g5-C5H5)Fe(CO)C(O)MePPhnPy3ꢀn
complexes proceeded smoothly, following the literature
procedure of (g5-C5H5)Fe(CO)C(O)MePPh3 [27] and gave
47–59% isolated yields. The PPhnPy3ꢀn analogs were char-
acterized with mp, IR, NMR, MS, and elemental analysis,
where the spectroscopic data are reasonable. The mCO
stretching frequencies of the acetyl complexes increase in
the following order, e.g., 1916 cmꢀ1 for the PPh3 complex,
1921 cmꢀ1 for the PPh2Py complex, 1923 cmꢀ1 for the
PPhPy2 complex, and 1927 cmꢀ1 for the PPy3 complex.
Clearly there is a decrease in backbonding from Fe to the
CO ligand with an increase of Py-groups, suggesting that
PPy3 is the least donating arylpyridylphosphine in the ser-
ies. On the other hand, the CV oxidative wave of the acetyl
complex shows a decrease in potential (0.244 V, 0.227irrev V,
0.200irrev V, and 0.195rrev V corresponding to 0, 1, 2, and 3
Py groups), indicative of PPy3 the best donating ligand in
the series, making (g5-C5H5)Fe(CO)C(O)MePPy3, the easi-
est to be oxidized. The mCO stretching frequencies and Eꢁ
values are in opposite directions! The pK value of
!
X
d
¼ 64:13d þ 3:25
ri
:
ðcation complexÞ
i
The calculated chemical shifts and the observed ones are
within 2.0 d. The ratio 3.25 is also suggestive of an
amplification. Upon complexation, the PPhnPy3ꢀn series
is apparently different from the PPhnMe3ꢀn series because
in the latter case, the linear regression for ligands
PPhnMe3ꢀn and the corresponding complexes (g4-exo-
MeC5H5)Fe(CO)2PPhnMe3ꢀn indicates
a
regulation,
instead of an amplification, about the phosphine group
contributions upon complexation where the ratio is only
0.85 that is less than 1.00 [28].
The mCO stretching frequencies of the ½ðg5-C5H5Þ-
FeðCOÞ2PPhnPy3ꢀnþꢂ series are not very much different
and are not conclusive in differentiating the PPhnPy3ꢀn
donor capability at all. Similar to the reported CV of
½ðg5-C5H5ÞFeðCOÞ2PPh þꢂ by Stweigart et al. [18], the cyc-
lic voltammograms of ½3ðg5-C5H5ÞFeðCOÞ2PPhnPy3ꢀnþꢂ in
our study gave irreversible, reductive waves (ꢀ1.234 V,
ꢀ1.353 V, ꢀ1.380 V, ꢀ1.414 V, in the order of 0, 1, 2, and
3 Py groups), indicative of a reactivity after being reduced.
It is then worth trying the phosphine-exchange experiments.
A 10 mm coaxial NMR tube has been employed that has an
inner-tube containing PPh3(O)/CDCl3 and the outer-tube
containing a CH3CN solution of the (g5-C5H5)Fe(CO)2-
complex and phosphines. For example, the 31P NMR spec-
trum for a 3:1:1:1 mixture of ½ðg5-C5H5ÞFeðCOÞ2PPh3þꢂ=
PPy3=PPhPy2=PPh2Py gave respective peaks at d 62.5,
þ
HPPhnPy3ꢀn is not a good indicator here because of pyr-
idyl-N complication to the P base site – the titration curves
of PPhnPy3ꢀn (n = 1–3) using an acid are hardly resolved. It
is not conclusive, based on one of the Giering’s parameters,
to specify the best donor ligand in the PPhnPy3ꢀn series.
The preparation of ðg5-C5H5ÞFeðCOÞ2PPhnPy3ꢀnþꢂ
½PF6þꢂ follows the ETC PPhnPy3ꢀn ligand replacement
for Iꢀ on (g5-C5H5)Fe(CO)2I. The yields are excellent (in
77–90% isolated yields) and the spectroscopic data are in
agreement with the structural formula.

O.G. Adeyemi, L.-K. Liu / Inorganica Chimica Acta 360 (2007) 2464–2470
2469
ꢀ0.6, ꢀ1.9, ꢀ3.2 and intensity integration ratios of ca.
3.1. Fp+-affinity for PPhnPy3ꢀn
19:6:6:6. After the addition of trace amount of (g5-
C5H5)2Co, an initiator to activate the 17e–19e pathwaþy,
The equilibrium constants in Table 1 clearly reveal that
for the series of arylpyridylphosphines PPhnPy3ꢀn, the
donor capability increases in the order PPh3 < PPh2Py
< PPhPy2 < PPy3. The standard deviations of Kij’s are
within 20% which seems reasonable in view of uncorrected
instrumentation factors. These equilibrium constants could
the mixture reveals a 31P NMR spectrum of FpPPy3
,
FpPPhPy2þ, FpPPh2Py+, FpPPh3þ, PPy3, PPhPy2, PPh2Py,
PPh3 with respective peaks at d 76.1, 74.0, 70.5, 62.5, ꢀ0.6,
ꢀ1.9, ꢀ3.2, ꢀ4.7 and intensity integration ratios of ca.
10.80:10.47:8.08:7.94:2.36:3.99:6.94:32.46,
respectively.
The change of the spectrum is both instantaneous and
extensive. The time delay between two spectra is less than
a few minutes, barely enough for one to introduce the
(g5-C5H5)2Co reductant to the 10 mm NMR tube and set
up a new NMR experiment. Four new species are formed
after addition of (g5-C5H5)2Co – the peak at d ꢀ4.7 is indic-
ative of released PPh3 and the three new resonances at d
76.1, 74.0, and 70.5 gave evidence of the formation of
½ðg5-C5H5ÞFeðCOÞ2PPy3þꢂ, ½ðg5-C5H5ÞFeðCOÞ2PPhPy2þꢂ,
and [(g5-C5H5)Fe(CO)2PPh2Py+], respectively. The inten-
sity integration ratios for ½ðg5-C5H5ÞFeðCOÞ2PPy3þꢂ,
Py+], ½ðg5-C5H5ÞFeðCOÞ2PPh3þꢂ, PPy3, PhPy2, PPh2Py,
and PPh3 (in decreasing order of d values) become ca.
11:10:8:8:2:4:7:32. The ratios stay the same after 24 h, i.e.,
the phosphine exchange reaction with a radical process
proceeds to a completion in a time scale of sub-minutes.
For a relevant pair of phosphines, the equilibrium con-
stant K of Eq. (1) could be calculated based on the intensity
integration ratio. Thus, out of 4 complexes and 4 ligands
involved after the addition of (g5-C5H5)2Co, there are 16
phosphine exchange equilibrium constants Kij (i, j = 1–4),
defined in Table 1, among which four are the self-exchange
(K11 = K22 = K33 = K44 = 1), six the forward reactions (Kij
with i < j) and six the reverse reactions (Kij with i > j). For
instance in the equation
further be deducted into a much more easily understood
Fp+-affinity scale where Fpþ ¼ ðg5-C5H5ÞFeðCOÞ
,
þ
2
which is a measure of the association between a PPhnPy3ꢀn
ligand and the Fp+ cation:
Fpþ þ PPh3 ¼ FpPPh3
K1 ¼ ½FpPPh3þꢂ=½Fpþꢂ½PPh3ꢂ
þ
K2 = [FpPPh2Py+]/[Fp+][PPh2Py]
Fp+ + PPh2Py = FpPPh2Py+þ
Fpþ þ PPhPy2 ¼ FpPPhPy2 K3 ¼ ½FpPPhPy2þꢂ=½Fpþꢂ½PPhPy2ꢂ
Fpþ þ PPy3 ¼ FpPPy3
K4 ¼ ½FpPPy3þꢂ=½Fpþꢂ½PPy3ꢂ
þ
Hence, the Kij’s listed in Table 1 could be factored, i.e.,
K12 = K2/K1, K13 = K3/K1, etc. As Fp+ is only loosely de-
fined and PPh3 is the most used triarylphosphine, K1 could
reasonably assume to be unitary in this Fp+-affinity scale
to quantify the parameter. Then K2 = 4.90, K3 = 11.0,
and K4 = 20.3.
½ðg5-C5H5ÞFeðCOÞ2PPhPy2þꢂ,
[(g5-C5H5)Fe(CO)2PPh2-
The incoming phosphine replacement for CO is gener-
ally more difficult than the replacement for phosphine in
view of a stronger Fe–C(CO) bond than Fe–P. In the
present radical phosphine exchange reactions, the
substitution by excess PPhnPy3ꢀn for CO was not
observed. We believe that the large cone angles of
PPhnPy3ꢀn are important here because in the 17e–19e
pathways, the reaction rate is very sensitive to steric hin-
drance. For phosphines of smaller cone angles, e.g.,
PPh2Me (H = 136ꢁ), we have observed that the phos-
phine exchange experiment starting PPh2Me and
½ðg5-C5H5ÞFeðCOÞ2PPh3þꢂ as initialized with a catalytic
amount of (g5-C5H5)2Co, yielded both [(g5-C5H5)-
ðg5-C5H5ÞFeðCOÞ2PPh3þ þ PPy3
þ
¼ PPh3 þ ðg5-C5H5ÞFeðCOÞ2PPy3
Fe(CO)2PPh2Me+] and ½ðg5-C5H5ÞFeðCOÞðPPh2MeÞ þꢂ
K14 ¼ ½ðg5-C5H5ÞFeðCOÞ2PPy3þꢂ½PPh3ꢂ
=½ðg5-C5H5ÞFeðCOÞ2PPh3þꢂ½PPy3ꢂ
ꢃ ð11Þð32Þ=ð8Þð2Þ ¼ 22:
2
as products [29]. The double replacement product could
be reduced at lower temperatures. A study is now under-
way in a hope to extend the Fp+-affinity scale to PPh2Me
and/or phosphines of smaller cone angles.
Listed in Table 1, the actual calculations have been per-
formed using the intensity integration values in two deci-
mal points (K14 is 18.7, instead of the approximate 22).
The rest of equilibrium constants are also calculated in this
manner. The conversion factors relating the concentration
of a PPhnPy3ꢀn or a ½ðg5-C5H5ÞFeðCOÞ2PPhnPy3ꢀnþꢂ and
the intensity integration in 31P NMR spectrum may be dif-
ferent. In our experience, these factors are about the same
if an error of 10% is taken into consideration. Experi-
ments with simultaneously present arylpyridylphosphines
indeed deliver a very nice internal consistency. The initial
concentrations with three portions of one ðg5-C5H5Þ-
4. Conclusion
Application of a radical process results in the rapid phos-
phine exchange between the ½ðg5-C5H5ÞFeðCOÞ2PAr3þꢂ
complexes and the PR3 ligands (PAr3, PR3 = PPhnPy3ꢀn
,
n = 0–3) where the donor capability increases in the order
PPh3 < PPh2Py < PPhPy2 < PPy3 with quantified Fp+ affin-
ity scales at 1, 4.90, 11.0, and 20.3, respectively.
Acknowledgements
ꢀ
FeðCOÞ2 phosphine complex and one portion each of
the remaining arylpyridylphosphines are listed in Table 1
that summarizes the phosphine exchange results.
Thanks are due to the National Science Council and
Academia Sinica for the kind financial support.

2470
O.G. Adeyemi, L.-K. Liu / Inorganica Chimica Acta 360 (2007) 2464–2470
(c) S. Xiaxo, W.C. Trogler, D.E. Ellis, Z. Berkovitch-Yellin, J. Am.
References
Chem. Soc. 105 (1983) 7033.
[11] (a) A.M. Bond, S.W. Carr, R. Colton, D.P. Kelly, Inorg. Chem. 22
(1983) 989;
[1] E.C. Alyea, D.W. Meek (Eds.), Catalytic Aspects of Metal Phosphine
Complexes, Advances in Chemistry Series 196, American Chemical
Society, Washington, DC, 1982.
[2] M.E. Cucullu, C. Li, S.P. Nolan, S.T. Nguyen, R.H. Grubbs,
Organometallics 17 (1998) 5565.
(b) D.J. Kuchynka, C. Amatore, J.K. Kochi, Inorg. Chem. 25 (1986)
4087;
(c) J.W. Hershberger, J.K. Kochi, Polyhedron 2 (1983) 929;
(d) C.W. Smith, G.W. Vanloon, M.C. Baird, J. Coord. Chem. 6
(1976) 89;
[3] C.A. Tolman, Chem. Rev. 77 (1977) 313.
[4] (a) C.A. Streuli, Anal. Chem. 31 (1959) 1652;
(b) W.A. Henderson, C.A. Streuli, J. Am. Chem. Soc. 82 (1960) 5791;
(c) T. Allman, R.G. Goel, Can. J. Chem. 60 (1982) 716;
(d) R.A. Jackson, R. Kanluen, A. Poe, Inorg. Chem. 23 (1984) 523.
[5] (a) T. Bartik, T. Himmler, H.-G. Schulte, K. Seevogel, J. Organomet.
Chem. 272 (1984) 29;
(e) N.G. Connelly, M.D. Kitchen, J. Chem. Soc., Dalton Trans.
(1977) 931.
[12] (a) A.G. Orpen, N.G. Connelly, J. Chem. Soc., Chem. Commun.
(1985) 1310;
(b) M.J. Wovkulich, J.L. Atwood, L. Canada, J.D. Atwood,
Organometallics 4 (1985) 867;
(c) F.A. Cotton, D.J. Darensbourg, S. Klein, B.W.S. Kolthammer,
Inorg. Chem. 21 (1982) 294;
(d) F.A. Cotton, D.J. Darensbourg, S. Klein, B.W.S. Kolthammer,
Inorg. Chem. 20 (1981) 2661;
(b) C.A. Tolman, J. Am. Chem. Soc. 92 (1970) 2953;
(c) R.G. Goel, W.P. Henry, R.C. Srivastava, Inorg. Chem. 20 (1981)
1727;
(d) G.M. Gray, C.S. Kraihanzel, J. Organomet. Chem. 241 (1983) 201;
(e) P.M. Treichel, Inorg. Chem. 7 (1968) 1942;
(e) N. Bresciani-Pahor, M. Forcolin, L.G. Marzilli, L. Randaccio,
M.F. Summers, P.J. Toscano, Coord. Chem. Rev. 63 (1985) 1;
(f) F.A. Cotton, D.J. Darensbourg, W.H. Ilsley, Inorg. Chem. 20
(1981) 578;
(g) E.C. Alyea, C. Ferguson, A. Somogyvari, Inorg. Chem. 21 (1982)
1369;
(f) W.A.G. Graham, Inorg. Chem. 7 (1968) 315;
(g) J.O. Albright, F.L. Tanzella, J.G. Verkade, J. Coord. Chem. 6
(1976) 225;
(h) O. Stelzer, E. Unger, Chem. Ber. 108 (1975) 1246;
(i) C. Barbeau, J. Turcotte, Can. J. Chem. 54 (1976) 1603;
(j) A. Kececi, D. Rehder, Z. Naturforsch. B 36 (1981) 20;
(k) R. Uson, V. Riera, J. Gimeno, M. Laguna, M.P. Gamasa, J. Chem.
Soc., Dalton Trans. (1979) 996;
(l) K. Ihmels, D. Redher, Organometallics 4 (1985) 1334;
(m) P.S. Braterman, D.W. Milne, E.W. Randall, E. Rosenberg, J.
Chem. Soc., Dalton Trans. (1973) 1027;
(n) B.E. Mann, C. Masters, B.L. Shaw, J. Chem. Soc. (1971) 1104;
(o) J.A. Timney, Inorg. Chem. 18 (1979) 2502;
(h) F.A. Cotton, D.J. Darensbourg, B.W.S. Kolthammer, Inorg.
Chem. 20 (1981) 4440.
[13] M. Rahman, Md.Y.-Y. Liu, K. Eriks, A. Prock, W.P. Giering,
Organometallics 8 (1989) 1.
[14] R.L. Rich, H. Taube, J. Am. Chem. Soc. 76 (1954) 2608.
[15] W.C. Trogler (Ed.), Organometqallic Radical Processes, Elsevier,
Amsterdam, 1990.
[16] (a) D.R. Tyler, Prog. Inorg. Chem. 36 (1988) 125;
(b) D. Astruc, Chem. Rev. 88 (1988) 1189;
(p) S. Vastag, B. Heil, L. Marko, J. Mol. Catal. 5 (1979) 189.
[6] (a) G.M. Bodner, C. Gagnon, D.N. Whittern, J. Organomet. Chem.
243 (1983) 305;
(c) M.C. Baird, Chem. Rev. 88 (1990) 1217;
(d) D.R. Tyler, F. Mao, Coord. Chem. Rev. 97 (1990) 119;
(e) D.R. Tyler, Acc. Chem. Rev. 24 (1991) 325.
[17] D. Astruc, Electron Transfer and Radical Process in Transition Metal
Chemistry, VCH, Weinheim, 1995.
[18] K.A. Pevear, M.M. Banaszak Holl, G.B. Carpenter, A.L. Rieger,
P.H. Rieger, D.A. Stweigart, Organometallics 14 (1995) 512.
[19] (a) M.Y. Darensbourg, J. Organomet. Chem. 38 (1972) 133;
(b) E.C. Ashby, T.N. Pham, B. Park, Tetrahedron Lett. 26 (1985)
4691;
(c) H. Yamataka, N. Fujimura, Y. Kawafujim, T. Hanafusa, J. Am.
Chem. Soc. 109 (1987) 4305.
[20] S.L. Gibson, L.-K. Liu, R.U. Soliz, J. Organomet. Chem. 526 (1996)
393.
(b) G.M. Bodner, M.P. May, L.E. Mckinney, Inorg. Chem. 19 (1980)
1951;
(c) G.M. Bodner, Inorg. Chem. 14 (1975) 2694;
(d) G.M. Bodner, Inorg. Chem. 13 (1974) 2563;
(e) G.M. Bodner, L.J. Todd, Inorg. Chem. 13 (1974) 1335;
(f) G.M. Bodner, Inorg. Chem. 14 (1975) 1932;
(g) S.O. Grim, R.M. Singer, J. Coord. Chem. 8 (1978) 121;
(h) A.F. Masters, G.E. Bossard, T.A. George, R.T.C. Brownlee,
M.J. O’Connor, A.G. Wedd, Inorg. Chem. 22 (1983) 908;
(i) R.J. Blau, J.H. Espenson, Inorg. Chem. 25 (1986) 878;
(j) R.V. Honeychuck, W.H. Hersh, Inorg. Chem. 26 (1987) 1826;
(k) T.T. Derencsenyi, Inorg. Chem. 20 (1981) 665;
(l) W.C. Trogler, L.G. Marzilli, Inorg. Chem. 14 (1975) 2942;
(m) E.C. Alyea, R.E. Lenkinski, A. Somogyvari, Polyhedron 1 (1982)
130;
[21] D.D. Perrin, W.L.F. Armarego, D.R. Perrin, Purification of Labo-
ratory Chemicals, Pergamon Press, Oxford, 1981.
[22] (a) B.D. Dombek, R.J. Angelici, Inorg. Chim. Acta 7 (1973) 345;
(b) T.J. Meyer, E.C. Johnson, N. Winterton, Inorg. Chem. 10 (1971)
1673;
(n) R.W. Balk, D.J. Stufkens, A. Oskam, Inorg. Chim. Acta 28
(1978) 133.
[7] F.A. Cotton, W.T. Edwards, F.C. Rauch, M.A. Graham, R.N.
Perutz, J.J. Turner, J. Coord. Chem. 2 (1973) 247.
[8] (a) B.E. Bursten, D.J. Darensbourg, C.E. Kellog, D.L. Lichtenberger,
Inorg. Chem. 23 (1984) 4361;
(c) T.J. Meyer, E.C. Johnson, N. Winterton, Inorg. Synth. 12 (1971)
36;
(d) T.J. Meyer, E.C. Johnson, N. Winterton, Inorg. Synth. 7 (1963)
110;
(b) J.W. Yarbrough, M.B. Hall, Inorg. Chem. 17 (1978) 2269;
(c) R.J. Puddephalt, L. Dignard-Bailey, G.M. Bancroft, Inorg. Chim.
Acta 96 (1985) L91;
(d) S. Ikuta, P. Kebarle, G.M.T. Bancroft, R.J. ChanPuddephatt, J.
Am. Chem. Soc. 104 (1982) 5899.
(e) R.B. King, M.B. Bisnette, J. Organomet. Chem. 2 (1964) 15.
[23] T.S. Piper, G. Wilkinson, J. Inorg. Nucl. Chem. 3 (1956) 104.
[24] A. Maisonnet, J.P. Farr, M.M. Olmstead, C.T. Hunt, Inorg. Chem.
21 (1982) 3961.
[25] (a) F.G. Mann, J. Watson, J. Org. Chem. 13 (1948) 502;
(b) K. Kurtev, D. Ribola, R.A. Jones, D.J. Cole-Hamilton, G.
Wilkinson, J. Chem. Soc., Dalton Trans. (1980) 56.
[26] P.M. Treichel, R.L. Shubkin, Inorg. Chem. 6 (1967) 1328.
[27] J.P. Bibler, A. Wojcicki, Inorg. Chem. 5 (1966) 889.
[28] L.-S. Luh, U.B. Eke, L.-K. Liu, Organometallics 14 (1995) 440.
[29] J. Peng, L.-K. Liu, Bull. Inst. Chem., Acad. Sin. 45 (1998) 27.
[9] (a) B.V. Johnson, A.L. Steinmetz, P.J. Ouseph, J. Coord. Chem. 14
(1985) 103;
(b) H. Inoue, M. Sasagawa, E. Fluck, Z. Naturforsch. B 40 (1985) 22;
(c) W.E. Carroll, F.A. Deeney, J.A. Delaney, F.J. Lalor, J. Chem.
Soc., Dalton Trans. (1973) 718.
[10] (a) D.S. Marynick, J. Am. Chem. Soc. 106 (1984) 4064;
(b) M.-H. Whangbo, K.R. Steward, Inorg. Chem. 21 (1982) 1720;