Synthetic, electrochemical and structural aspects of a series of ferrocene-containing dicarbonyl β-diketonato rhodium(I) complexes

Article abstract of DOI:10.1016/j.ica.2005.02.010

Treatment of [Rh(β-diketonato)(cod)] with CO resulted in better yields of [Rh(FcCOCHCOR)(CO)₂] than by treating [Rh(Cl)(CO) ₂]₂ with FcCOCH₂COR, R = CF₃ (Hfctfa), CH₃ (Hfca), Ph (Hbfcm, Ph

Full text of DOI:10.1016/j.ica.2005.02.010

Inorganica Chimica Acta 358 (2005) 2530–2542
www.elsevier.com/locate/ica
Synthetic, electrochemical and structural aspects of a series of
ferrocene-containing dicarbonyl b-diketonato rhodium(I) complexes
a
Jeanet Conradie , T. Stanley Cameron , Manuel A. S. Aquino ,
b
c
a
Gert J. Lamprecht , Jannie C. Swarts
a,
*
a
Department of Chemistry, University of the Free State, P.O. Box 339, Nelson Mandela Drive, Bloemfontein 9300, South Africa
b
Dalhousie University, Halifax, Nova Scotia, Canada B3H 4J3
c
St. Francis Xavier University, P.O. Box 5000, Antigonish, Nova Scotia, Canada B2G 2W5
Received 17 January 2005; accepted 6 February 2005
Available online 14 March 2005
Abstract
Treatment of [Rh(b-diketonato)(cod)] with CO resulted in better yields of [Rh(FcCOCHCOR)(CO)2] than by treating
[Rh(Cl)(CO)2]2 with FcCOCH2COR, R = CF3 (Hfctfa), CH3 (Hfca), Ph (Hbfcm, Ph = phenyl) and Fc (Hdfcm, Fc = ferrocenyl).
˚
The single crystal structure of the fctfa rhodium(I) complex [C16H10F3FeO4Rh], monoclinic, C 2/c(15), a = 13.266(3) A,
˚
˚
b = 19.553(3) A, c = 13.278(3) A, b = 100.92(2)ꢀ, Z = 8 showed both rotational and translational displacement disorders for the
0
0
CF3 group. An electrochemical study revealed that the formal reduction potential, E , for the electrochemically reversible one elec-
tron oxidation of the ferrocenyl group varied between 0.304 (for the fctfa complex) and 0.172 V (for the dfcm complex) versus Fc/
Fc+ in a manner that could be directly traced to the group electronegativities, vR, of the R groups on the b-diketonato ligands, as
well as to the pKa0 values of the free b-diketones. Anodic peak potentials, Epa,Rh, for the dominant cyclic voltammetry peak asso-
ciated with rhodium(I) oxidation were between 0.718 (bfcm complex) and 1.022 V (dfcm complex) versus Fc/Fc+. Coulometric
experiments implicated a second, much less pronounced anodic wave for the apparent two-electron RhI oxidation that overlaps with
the ferrocenyl anodic wave and that the redox processes associated with these two RhI oxidation waves are in slow equilibrium with
each other.
ꢁ 2005 Elsevier B.V. All rights reserved.
Keywords: Ferrocene b-diketones; Rhodium dicarbonyls; Electrochemistry; Crystal structure
1. Introduction
thoroughly studied [2,3]. In contrast, studies on com-
plexes of the type [Rh(b-diketonato)(CO)2] are rare.
No electrochemical properties relating to metal-con-
taining b-diketonato-dicarbonyl rhodium complexes
have to our knowledge been published. Only one ki-
netic study in which cyclooctadiene (cod) was used to
substitute the two carbonyl ligands [4] of [Rh(b-diketo-
nato)(CO)2] complexes, has been documented. Hickey
and Maitlis [5] reported on the kinetics of oxidative
addition of methyl iodide to the related dicarbonyl
[Rh(CO)2(I)2]ꢀ complex and Borrini and Ingrosso [6]
reported on the reaction between allene and [Rh(b-
diketonato)(CO)2] complexes.
Rhodium(I) dicarbonyl complexes were made fa-
mous by the use of [Rh(CO)2(I)2]ꢀ as catalyst in the
Monsanto process of converting methanol to acetic
acid [1]. Much less is known about b-diketonato-dicar-
bonyl complexes although the kinetics of oxidative
addition and b-diketonato substitution in complexes
of the type [Rh(b-diketonato)(CO)(PPh3)] have been
*
Corresponding author. Tel.: +2751 4012781; fax: +2751 4307805.
E-mail address: swartsjc.sci@mail.uovs.ac.za (J.C. Swarts).
0020-1693/$ - see front matter ꢁ 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.ica.2005.02.010

J. Conradie et al. / Inorganica Chimica Acta 358 (2005) 2530–2542
2531
R
2. Results and discussion
O
O
[Rh(Cl)(cod)]2
2.1. Cyclic voltammetry
Rh
Red [Rh(b-diketonato)(CO)2] complexes with b-dike-
tonato = fctfa (9), fca (10), bfcm (11) and dfcm (12) were
obtained by two different pathways (Scheme 1) in yields
up to 80% following the general guidelines of published
procedures [13]. The complexes 9–12 have two redox
centres, the iron centre of the sandwich ferrocenyl group
and the square planar rhodium(I) centre.
The ferrocenyl (Fc) group is seemingly oxidised to the
ferrocenium species at less positive potentials than the
rhodium centre (but see later). CV experiments in a po-
tential range where rhodium(I) did not exhibit any obvi-
ous redox activity indicated reversible electrochemical
behaviour for the ferrocenyl group in compounds 9–11
with DEp = Epa ꢀ Epc 6 90 mV at scan rates up to
250 mV sꢀ1 in acetonitrile (Table 1, Fig. 1). In compar-
O
O
Fe
R = CF3 (5), CH3 (6),
Ph (7), Fc (8)
R
Fe
CO(g)
R
O
O
CO
CO
1: R = CF3
2: R = CH3
3: R = C6H5
4: R = Fc
[Rh(Cl)(CO)2]2
Rh
Fe
R = CF3 (9), CH3 (10),
Ph (11), Fc (12)
Scheme 1. The synthesis of the [Rh(b-diketonato)(CO)2] complexes 9–
12 were achieved in higher yields by cyclooctadiene displacement from
5–8 with CO than by direct treatment of the dimer [Rh(Cl)(CO)2]2 with
b-diketones 1–4. Fc = ferrocenyl, Ph = phenyl.
ison, we found that for ferrocene itself DEp =
+
74 mV, ipc/ipa = 0.98, E0 = 0.087 V versus Ag/Ag (as
0
0.01 mol dmꢀ3 AgNO3 in CH3CN). Determinable peak
cathodic/peak anodic current ratios for 9–12 were in
the range 0.83 < ipc/ipa < 1 and the formal reduction
The development of new materials [7] possessing mul-
tiple metal centres are lately receiving an increasing
amount of attention in applications such as molecular
wires [8], catalysis [9] and luminescence-based sensors
[10]. In many of these cases, the iron-containing ferroce-
nyl fragment is employed as an electron reservoir [7]. We
recently reported [11] the synthesis of a series of new fer-
rocene-containing b-diketones including 1–4 (Scheme 1)
and their rhodium–cod complexes 5–8, as well as the
kinetics [12] of b-diketonato substitution of 5–8 with
1,10-phenathroline. We now describe a further portion
of our ongoing investigation into the chemistry of rho-
dium–ferrocene complexes that specifically focuses on
the structural and electrochemical properties of the fer-
rocene-containing b-diketonato dicarbonyl rhodium
complexes 9–12 (Scheme 1).
potentials, E0 = 1/2(Epa + Epc), were scan rate indepen-
0
dent between 50 and 250 mV sꢀ1. Complex 12, having
two ferrocenyl groups in its structure, exhibited two
poorly resolved ferrocene-based waves (Fig. 1).
Although the free ligand Hdfcm has two spectroscopi-
cally equivalent ferrocenyl groups (as indicated by the
1H NMR spectrum), they are not oxidised simulta-
neously during cyclic voltammetry [14]. E0 values for
0
Hdfcm were reported as 0.178 and 0.287 V versus Fc/
Fc+ (after correcting for the reference electrode Ag/
Ag+ by subtracting 00.087 V). For complex 12, the two
ferrocenyl-based E0 values are estimated to be 0.172
and 0.277 V versus Fc/Fc+, respectively. The smaller for-
mal reduction potential of these two is also the least po-
Table 1
Cyclic voltammetry and coulometry data of the ferrocenyl (Fc) moiety for 1 mmol dmꢀ3 solutions of [Rh(FcCOCHCOR)(CO)2] in acetonitrile
containing 0.1 mol dmꢀ3 (nBu)4NPF6 at 25 ꢀC and a scan rate of 50 mV sꢀ1
ipc;Fc
ipa;Fc
a
vR
b
Complex (R)
pK0a
Epa, Fc (V)c
ipa,Fc (lA)
DEp,Fc (mV)
E0 (V)c
0
Ecoul (V)cd
ne
9 (CF3)
10 (CH3)
11 (C6H5)
12 (Fc)
a
3.01
2.34
5.65
10.01
0.345, (0.457)f
0.224
0.235
21.68, (–)f
18.82
20.31
82, ( 74)f
68
72
–, (–)f
0.85
0.304, ( 0.421)f
0.190
0.199
0.713
0.563
0.513
0.513
3
3
3
4
2.21
1.87, (2.82)g
10.41
13.1, (6.80)h
0.86
1.00, (0.83)i
0.207, (0.312)i
18.2, (18.6)i
69, (70)i
0.172, (0.277)i
Group electronegativities of the indicated R-groups [14].
b
c
d
e
f
For the free b-diketone, FcCOCH2COR [12,14]. A pK0a of 6.53 is also reported [14] for the CF3 b-diketone.
All potentials are vs. the Fc/Fc+ couple.
Coulometry experiments were performed at Epa,Fc, as well as at Ecoul
Total amount of electrons per molecule that were counted during the oxidation of 9–12 at Epa,Fc as well as at Ecoul
The second data set is associated with the ferrocenyl peak in [Rhoxidised(FcCOCHCOCF3)(CO)2(solvent)x] and manifests as a shoulder to the main
.
.
ferrocenyl wave. Poor resolution of these two peaks prevented accurate determination of data other than Epa
g
.
The value in brackets is the group electronegativity of the oxidised ferrocenium species, Fc+.
pK0a of Fc+COCH2COFc.
The first data entry is associated with electron transfer of the first ferrocenyl group, the second with that of the second ferrocenyl group.
h
i

2532
J. Conradie et al. / Inorganica Chimica Acta 358 (2005) 2530–2542
reduction potentials of the Fc groups in 9–12 is
132 mV, which is in the same order than that observed
for the free b-diketones (129 mV) [14].
[Rh(fctfa)(CO)2]
[Rh(bfcm)(CO)2]
25 µA
The cyclic voltammogram of the ferrocenyl group of
9, [Rh(FcCOCHCOCF3)(CO)2], introduced a phenome-
non that probably also exists for complexes 10–12, but
was not measurable under our experimental conditions.
Just one peak was expected for the ferrocenyl-based
electron transfer process but two poorly resolved peaks
were clearly detected in the potential range
0.0 < E < 0.7 V, Fig. 2. OSWV also shows these two
peaks at 0.289 and 0.419 V versus Fc/Fc+. To explain
this dual peak observation, comparison of the 1H
NMR spectrum of this compound in CDCl3 and
CD3CN, Fig. 2, is instructive. The elemental analysis
[13] of compounds 9–12 showed them to be pure. The
1H NMR spectrum obtained from poorly co-ordinating
CDCl3 solutions is, therefore, that of pure 9 and the
crystal structure determination of 9 is discussed at the
end of this publication. However, with deuterated aceto-
nitrile as solvent, the existence of a second species along-
side the main species is clearly observed. At 10 mmol
dmꢀ3, the minor species is thought to be an acetonitrile
adduct of 9, viz. [Rh(FcCOCHCOCF3)(CO)2(CD3CN)]
while the main component in solution is considered as
‘‘free’’ 9. Rhodium(I) complexes do favour a four co-
ordinate 16 electron square-planar co-ordination sphere
[19], but 18-electron five co-ordinated trigonal bipyrimi-
dal complexes have been isolated [20,21]. The possibility
also exists that the additional set of peaks observed in
the CD3CN spectrum of 9 is that of a second rhodium
species in the sample and that the use of CD3CN rather
than CDCl3 simply provided sufficient additional shift
dispersion to permit the minor species to be observed.
To eliminate this possibility in favour of the equilibrium
[Rh(fca)(CO)2]
[Rh(dfcm)(CO)2]
20 µA
0.0
0.2
0.4
0.80
-0.20
0.30
E / V vs. Fc/Fc+
Fig. 1. Cyclic voltammetry of 1 mmol dmꢀ3 solutions of complexes in
acetonitrile at 25 ꢀC utilising a glassy carbon working electrode. Scan
rate = 100 mV sꢀ1, supporting electrolyte = 0.1 mol dmꢀ3 (nBu)4NPF6.
Inset bottom right. Osteryoung square wave voltammetry (scan
rate = 15 mV sꢀ1) clearly indicates two inequivalent electron transfer
processes for the two ferrocenyl groups of [Rh(dfcm)(CO)2].
sitive formal reduction potential observed in the com-
pound series 9–12. Osteryoung square wave voltamme-
try (OSWV) more clearly demonstrated the
electrochemical inequivalence of the two ferrocenyl
groups of 12 with peak maxima at 157 and 259 mV,
respectively (inset bottom right, Fig. 1). To understand
the electrochemical inequivalence of the two ferrocenyl
groups of [Rh(FcCOCHCOFc)(CO)2], one should rec-
ognise that the first oxidised intermediate species is the
mixed-valent species [Rh(Fc+COCHCOFc)(CO)2] simi-
larly to what was found for the free b-diketone [14].
The inequivalence of the ferrocenyl and ferrocenium
groups of this intermediate is highlighted in terms of
½RhðFcCOCHCOCF3ÞðCOÞ ꢁ þ CD3CN
2
Kc0
ꢀ½RhðFcCOCHCOCF3ÞðCOÞ ðCD3CNÞꢁ
ð1Þ
2
their group electronegativities [11,14]:
þ
v
Fc = 1.87,
one should recognise that all equilibria are temperat0ure
dependent according to the equation ln K0c2 ¼ ln Kc1ꢀ
ðDH=RÞðT02 ꢀ T 10 Þ with Kc0 1 and Kc0 2 the observed equilib-
rium constants at temperatures T2 and T1, R =
8.314 J Kꢀ1 molꢀ1 and DH the reaction enthalpy [22].
Temperature dependent 1H NMR of a sample of 9 in ace-
tonitrile showed that K0c applied to the above equilibrium
changed from K0c½CD3CNꢁ ¼ ½RhðFcCOCHCOCF3Þ-
ðCOÞ ðCD3CNÞꢁ=½RhðFcCOCHCOCF3ÞðCOÞ ꢁ ¼ ðinte-
vFc ¼ 2:82. The electron withdrawing power of the ferr-
ocenium group is almost as high as that of the CF3
group ðvCF ¼ 3:01Þ. As there is good communication
3
(via conjugation) between the Fc and Fc+ b-diketonato
pendent side groups on the pseudo-aromatic core of the
intermediate [Rh(Fc+COCHCOFc)(CO)2], oxidation of
the second Fc group to yield the final oxidised species
[Rh(Fc+COCHCOFc+)(CO)2] takes place at a more po-
sitive potential than observed for the oxidation of the
first Fc group. Different formal reduction potentials
for side groups on symmetrical complexes in which
mixed-valent intermediates are generated are well
known in systems that allow electron delocalisation,
either through bridge mediated paths or from a direct
metal–metal interaction [14–18]. The spread in formal
2
2
gral of the 2H peak at 4:80 ppmÞ=ðintegral of the 2H
peak at 4:75 ppmÞ varies from 0.11 at 35 ꢀC to 0.082 at
60 ꢀC. One can also expect that the position of the above
equilibrium should be [Rh(FcCOCHCOCF3)(CO)2] con-
centration dependent. At 20 ꢀC, the value of K0c½CD3CNꢁ
changed as follows (values in brackets are the initially
weighed concentration [Rh(FcCOCHCOCF3)(CO)2] in

J. Conradie et al. / Inorganica Chimica Acta 358 (2005) 2530–2542
2533
Fig. 2. Left. CVs of solutions of 9 at scan rates of 50, 100, 150 and 200 mV sꢀ1, with conditions as per Fig. 1, show two ferrocenyl waves. Peak 1
represents the ferrocenyl-based redox process associated with of [Rh(FcCOCHCOCF3)(CO)2(CH3CN)] (9b) while peak 2 is associated with the
ferrocenyl group of [Rhoxidised(FcCOCHCOCF3)(CO)2(solvent)x]. Inset. OSWV analysis confirmed the presence of peaks 1 and 2 at 0.289 and
0.419 V, respectively. Scan rate = 4 mV sꢀ1. Right. The 1H NMR spectrum in the region 4–6.5 ppm of 9 in CD3CN (bottom) clearly shows the
existence of the adduct [Rh(FcCOCHCOCF3)(CO)2(CD3CN)] (9b) alongside that of the parent compound 9. In weakly co-ordinating CDCl3 (top),
only one species, free 9, is observable. Assignments (CD3CN): 1H NMR: 4.25 (s, C5H5 of 9, 5H), 4.30 (s, C5H5 of 9b, 5H), 4.76 (t, C5H4 of 9, 2H),
4.79 (t, C5H4 of 9b, 2H), 4.99 (m, C5H4 of 9b, 2H, overlapping with C5H4 of 9, 2H), 6.350 (s, COCHCO of 9b, 1H), 6.364 (s, COCHCO of 9, 1H).
mmol dmꢀ3): 0.14 (15.2); 0.18 (9.35); 0.28 (5.79); 0.55
(2.34); 5.4 (0.93). The observed trend is in agreement with
the expectation that equilibrium 1 should be driven more
quantitatively to the right when the relative amount of
CH3CN in a sample is increased over that of [Rh(FcC-
OCHCOCF3)(CO)2]. It is also instructive to realise that
observation of the acetonitrile adducts, such as [Rh(FcC-
OCHCOCF3- (CO)2(CD3CN)], is not a unique occur-
rence in transition metal chemistry, although it is scarce.
CH3CN adducts were also observed [23], for example,
in copper/phenanthroline complexes of the type [Cu(o-
phen)2(CH3CN)](BF4)2 with o-phen various substituted
1,10-phenanthroline derivatives. We conclude that the
reason why the above equilibrium is not forced com-
pletely to the right under the conditions of our cyclic vol-
tammetry experiments is a consequence of the smallness
of the K0c value in the product Kc0 ½CD3CNꢁ. However, it
is not so small that the equilibrium condition can be trea-
ted independent of the solvent because K0c is [Rh(FcC-
OCHCOCF3)(CO)2] concentration dependent. Hence
we prefer to present the apparent equilibrium constant
as obtained from 1H NMR data not as K0c, but rather as
the product K0c½CD3CNꢁ. It should be stressed that the
above sited values for K0c½CD3CNꢁ does not represent a
true thermodynamic equilibrium constant which amongst
others also implies compound activity coefficient = 1, but
merely represents the observed equilibrium constant
under the physical conditions of each experiment.
the adduct, [Rh(FcCOCHCOCF3)(CO)2(CD3CN)].
Since our electrochemistry was performed in CH3CN
containing 1 mmol dmꢀ3 total rhodium, peak 1 in the
CV of the solution containing [Rh(FcCOCH-
COCF3)(CO)2] (Fig. 2) is assigned to the ferrocenyl
group of the [Rh(FcCOCHCOCF3)(CO)2(CH3CN)]
adduct. Peak 2 is thought to be associated with the
ferrocenyl wave of either solvent-free [Rh(FcCOCH-
COCF3)-(CO)2] or, more consistent with our bulk elec-
trolyses results which are discussed below, with the
ferrocenyl wave of the rhodium-oxidised species [Rhoxi-
dised(FcCOCHCOCF3)(CO)2(solvent)x], Rhoxidised is
either RhII or RhIII. In principle, peak 2 could also rep-
resent minute traces of rhodium(I) oxidation, but rho-
dium oxidation is not electrochemically or chemically
reversible. A clear reduction wave for peak 2 is, how-
ever, apparent in the cathodic sweep (Fig. 2) at Epa
approximately 383 mV. At a scan rate of 50 mV sꢀ1 it
follows that DEp = Epa ꢀ Epc ꢂ 74 mV for peak 2. This
value is in good agreement with the theoretical value
of 59 mV for one electron transfer electrochemical
reversible processes, and with the DEp value which we
obtained under our experimental conditions for free fer-
rocene itself. An obvious feature of peak 2 is that it be-
comes less prominant than peak 1 at faster scan rates.
This observation will be explained after the oxidation
of rhodium(I) has been discussed.
Having determined that peak 1 in the CV of 9 (Figs. 1
and 2) actually corresponds to the species [Rh(FcC-
OCHCOCF3)(CO)2(CH3CN)], the question now arises
as to whether the single ferrocenyl peak in the CV of
From the equilibrium constants sited above, it fol-
lows that in CD3CN, at 0.93 mmol dmꢀ3, approximately
84% of the total rhodium species in solution has to be

2534
J. Conradie et al. / Inorganica Chimica Acta 358 (2005) 2530–2542
pseudo-aromatic core of these complexes exists. Fur-
14
11
Hdfcm
thermore, this linear relationship would be very fortu-
itous indeed if 10–12 were entirely in the free form
whileꢃ0 9 is in the adduct form. The linearity of the
vR=EFc relationship is regarded as consistent with the
view that 10–12, under the conditions of this electro-
chemical study, are also in the adduct form:
[Rh(FcCOCHCOR)(CO)2(CH3CN)].
Hbfcm
Hfca
8
Hdfcm+
In contrast to the ferrocenyl group, the rhodium
core of complexes 9–12 does not exhibit reversible
behaviour at all. Anodic (oxidation) peak potentials
of these complexes varied between 0.718 V, for the
bfcm complex 11, and 1.022 V for the dfcm complex
12, Table 2 and Fig. 4. In all the complexes 9–12, when
the rhodium(I) nucleus was oxidised at these potentials,
the ferrocenyl fragments of the molecule were already
in the ferrocenium oxidation state. The spread in peak
rhodium anodic potentials is 304 mV, more than dou-
ble that observed for the ferrocenyl fragment
(132 mV). This increase is partly attributable to the
large amount of electrochemical work that has to be
performed in going from a tetra co-ordinated square
planar dicarbonyl rhodium(I) reactant complex, either
to an oxidised RhIII complex, which normally has octa-
hedral co-ordination geometry [24], or to a RhII spe-
cies. If RhII is formed, a co-ordination sphere change
also has to take place. In contrast, the co-ordination
sphere of the ferrocenyl fragment does not change dur-
ing oxidation or reduction. The nature of the addi-
tional new ligands of the electrochemically generated
oxidised rhodium complexes during the course of the
cyclic voltammetry studies is not known, but probably
are solvent molecules, here acetonitrile (donor num-
ber = 14.1, see ref. [25,26]).
Hfctfa
CF3
5
3
Fc+
CH3
2.5
C6H5
2
Fc
1.5
100
300
E/Vvs. Fc/Fc+
Fig. 3. Bottom. A linear relationship between group electronegativi-
ties, vR, of each R group and E0 (ferrocenyl group) of [Rh(FcC-
OCHCOR)(CO)2(CH3CN)] demonstrates good communication
between the ferrocenyl group and each of the R-groups CF3, Fc+,
Ph, CH3 and Fc on the b-diketonato ligand. The pK0a of 6.53 (marked
with d, see ref. [14]) was ignored as it does not find the trend line. Top.
0
0
0
The formal reduction potential, E , of the ferrocenyl group of
[Rh(FcCOCHCOR)(CO)2(CH3CN)] is inversely proportional to pKa0
of the free b-diketone FcCOCH2COR which also confirms good
communication via the pseudo-aromatic b-diketonato core between
the ferrocenyl group and R-groups pendent to the b-diketonato ligand.
10–12 are those of the ‘‘free’’ complex or of the acetoni-
trile adduct. To decide which one, the linear relationship
between group electronegativity, vR, of each R-group of
For the reduction of the newly generated oxidised
rhodium species, two separate cathodic waves were ob-
served in the potential range ꢀ1.2 < Epc,Rh < ꢀ0.7 V
versus Fc/Fc+ (Table 2, Fig. 5). That these small catho-
dic waves (for example, for the fca complex, 10, ipc/
ipa ꢄ 0.16 at a scan rate of 50 mV sꢀ1, becoming slightly
00
9–12 (Table 1) and E (Fig. 3) is instructive. This linear
Fc
relationship clearly demonstrates that very good com-
munication between the ferrocenyl group and each of
the R-groups on [Rh(FcCOCHCOR)(CO)2] via the
Table 2
Cyclic voltammetry and coulometry data relating to the RhI nucleus in [Rh(FcCOCHCOR)(CO)2] complexes in acetonitrile containing 0.1 mol dmꢀ3
(nBu)4NPF6 at 25 ꢀC and a scan rate of 100 mV sꢀ1
N (mol)c
Q/C
Q/NF
Nd
a
ab
Complex (R)
Epa,Rh
ipa,Rh
Epc,Rh
ipc,Rh
9 (CF3)
10 (CH3)
11 (C6H5)
12 (Fc)
a
0.901
0.844
0.718
1.022
6.5
ꢀ0.687, (ꢀ1.091)e
ꢀ0.920, (ꢀ1.187)e
ꢀ0.831, (ꢀ1.113)e
ꢀ0.82f, (ꢀ0.99)f
12.7, (3.5)g
4.1, (2.5)g
4.9, (3.7)g
6.9f, (ꢀ)f
5.2 · 10ꢀ6
6.3 · 10ꢀ6
6.8 · 10ꢀ6
4.0 · 10ꢀ6
1.46
1.78
1.99
1.53
2.91
2.93
3.03
3.96
3
3
3
4
18.2
16.3
13.5
All potentials are vs. Fc/Fc+.
b
c
Within the potential window possible for acetonitrile, two RhIII reduction peaks are observed for 9–12.
Amount of moles of [Rh(FcCOCHCOR)(CO)2] used during coulometric experiments.
d
Total amount of electrons per [Rh(FcCOCHCOR)(CO)2] molecule that were counted during oxidation of 9–12 at an applied coulometric
potential of Epa,Rh, as well as 1113 mV.
e
Value in brackets are the more negative of two peak cathodic potentials associated with RhIII reduction in 9–12.
Poor resolution between cathodic peaks for this compound made meaningful measurements impossible. The indicated values are only estimates.
Current in brackets are associated with the more negative of the two Epc,Rh waves.
f
g

J. Conradie et al. / Inorganica Chimica Acta 358 (2005) 2530–2542
2535
dium species could electrochemically be generated at
potentials Epa,Rh > 0.7 V. The ferrocenyl group was,
however, oxidised. As demonstrated for complex 10 in
Fig. 5, these experiments clearly showed the ferrocenyl
reduction at 0.156 V and one weak cathodic peak at
ꢀ1.187 V. The peak at ꢀ0.920 V (complex 10) is, how-
ever absent. It follows that the peak at ꢀ0.920 V must
be associated with reduction of the rhodium species that
forms during the oxidation wave at 0.844 V. The source
of the reducing wave at ꢀ1.187 V is assigned to reduc-
tion of an oxidised rhodium species that was generated
in small quantities at a potential overlapping with the
peak associated with ferrocenyl oxidation (peak 1) in
the anodic sweep. Compelling proof that small quanti-
ties of RhIare oxidised when the ferrocenyl group is oxi-
dised will be presented when coulometric experiments
are discussed. The described experiments clearly indicate
that the reduction peaks between ꢀ1.2 and ꢀ0.7 V are
associated with the rhodium centre of 9–12.
210
150
90
[Rh(fctfa)(CO)2]
[Rh(bfcm)(CO)2]
[Rh(fca)(CO)2]
[Rh(dfcm)(CO)2]
30
0.14 0.34
-0.05
0.45
-30
-1.4
-0.4
0.6
1.6
E/V vs.Fc/Fc+
Fig. 4. CVs of [Rh(FcCOCHCOCF3)(CO)2] solutions, conditions are
as per Fig. 1. RhI oxidation is at Epa,Rh > 0.700 V, reversible
ferrocenyl-based redox processes are at 0.172 V < E0 < 0.421 V while
0
To get an indication of whether rhodium(I) is oxi-
dised to a rhodium(II) or a rhodium(III) species, a cou-
lometric study was performed. In these experiments,
electron counting was performed to see how many elec-
trons were transferred per molecule for each anodic
wave observed in the CV of each compound. When bulk
electrolyses was performed at potentials 200–300 mV
less positive than Epa,Fc, (Table 1), no electrons were
counted. When the potential of electrolyses were set to
be just at the onset of the ferrocenyl anodic wave, elec-
tron counts were so small (<0.4 after 6 h of electrolyses)
that they were not regarded as meaningful due to a lack
of electrochemical driving force. A surprising result was,
however, obtained during bulk electrolyses when a po-
tential equal to Epa,Fc was applied to solutions contain-
ing 9–12. This potential is sufficiently positive to oxidise
the ferrocenyl substituent but is significantly negative of
RhIII reduction was observed as two weak cathodic waves at
Epc,Rh < ꢀ0.7 V. The insets are enlargements of the ferrocenyl wave
of [Rh(dfcm)(CO)2] (12).
larger at faster scan rates: ipc/ipa ꢄ 0.28 at 250 mV sꢀ1
)
correspond to rhodium reduction were confirmed in
two ways. Firstly, in experiments where scans were initi-
ated at ꢀ0.287 V in a negative direction, the cathodic
peaks were absent. Fig. 5 demonstrates this for 10. It fol-
lows that the peaks at ꢀ1.2 to ꢀ0.7 V are not artefacts
of the experimental set-up (e.g., impurities in the sol-
vent). Rather, they must be an integral part of the
CVs of 9–12. In the second approach, a switching poten-
tial of 0.413 V versus Fc/Fc+ for scans in the positive
direction was imposed on the system. This meant that
the cathodic sweep was initiated before the oxidised rho-
Fig. 5. Left. CVs of solutions of [Rh(fca)(CO)2] (10) at a scan rate of 100 mV sꢀ1, with conditions as per Fig. 1. (a) When scanning commenced at
ꢀ287 mV in the negative direction (---) no cathodic reducing waves corresponding to RhIII reduction were observed. (b) When scanning was initiated
in a positive direction and a switching potential of 0.413 V imposed on the system (. . .), RhIII reduction at ꢀ1.187 V was observed. (c) When positive
initiated scans are subjected to a switching potential of 1.113 V, (—), two RhIII reductions (at ꢀ0.920 and ꢀ1.187 V) are observed. Right. CV of
solutions of [Rh(fca)(CO)2] (10) at scan rates of 50, 100, 150, 200 and 250 mVꢀ1. The ratio ipa,Rh/ipa,Fc becomes smaller with slower scan rates
indicating RhI oxidation at 0.844 V becomes less prominent while RhI oxidation occurring at ca. 0.2 V increases in intensity.

2536
J. Conradie et al. / Inorganica Chimica Acta 358 (2005) 2530–2542
the peak anodic potential of the rhodium centre (Table
2). It was expected to count one electron for the ferroce-
nyl nucleus of 9–11 and two electrons for 12. In practice,
three electrons were counted for 9–11 and four for 12.
This result is consistent with the simultaneous formation
of rhodium(III) on bulk electrolyses time scales. Three
(or four in the case of 12) electrons were also counted
when the applied potentials for couloumetric analysis
were at the deepest point of the trough between the
two anodic waves shown in Fig. 4 (values for these
potentials are indicated as Ecoul in Table 1). In separate
experiments, when the applied potential was Epa,Rh of
the rhodium-based anodic wave (Table 2), or 1113 mV
(i.e., about 200 mV larger than Epa,Rh), three and four
electrons were counted each time for 9–11 and 12,
respectively. Fig. 6 (left) shows the net amount of elec-
trons transferred per molecule of indicated complex,
n = Q/NF, that was obtained during these bulk electrol-
yses experiments.
This unexpected result led to two conclusions. For
the first, from the reversible nature of the ferrocenyl
couple, it is clear that each ferrocenyl moiety is involved
in a reversible one-electron transfer process, while the
rhodium nucleus most probably is involved in an irre-
versible two electron transfer process to possibly gener-
ate rhodium(III) during oxidation on the bulk
electrolyses time scale. For the second, to explain why
rhodium(III) apparently can be generated quantitatively
by electrolyses at potentials substantially less than
Epa,Rh, i.e., at Epa,Fc, one explanation would involve
the chemically irreversible nature of the rhodium oxida-
tion wave. The Nernst equation has a log term in it and,
no matter what the value of the electrode potential,
there should always be at least a small amount of each
half of the rhodium redox couple present in solution.
The peak anodic potential where rhodium(I) is oxidised
is where the redox reaction occurs at the highest rate.
Given the chemically irreversible nature of the rhodium
oxidation wave, it is possible that oxidation of the rho-
dium centre occurs at potentials significantly more neg-
ative than the peak potential. However, Epa,Fc is so
much smaller than Epa,Rh (from 0.483 V for 11 till
0.814 V for 12), it is difficult to explain the high rate
of rhodium electrolyses from this argument. We con-
clude that, apart from Epa,Rh > 0.7 V, there must exist
a second means of generating rhodium(III) at potentials
overlapping with the ferrocenyl anodic wave. Whatever
the means are by which the two electron transfer process
to possibly generate rhodium(III) at potentials overlap-
ping closely with Eꢀ0 of the ferrocenyl group, this redox
process can only occur to the extent that it draws small
currents because ipc/ipa ratios for the ferrocenyl wave
does not deviate substantially from one (Table 1). This
low-potential process must, however, be in slow equilib-
rium with the process observed at Epa,Rh (Table 2) to ex-
plain why a full electron count of three electrons
transferred per molecule of 9–11 or four electrons trans-
ferred for 12 during coulometric experiments is observed
at potentials substantially less positive than Epa,Rh. Fur-
ther proof for this slow equilibrium was found in ipa,Rh
/
ipa,Fc ratios. We were able to demonstrate that ipa,Rh de-
creased faster than ipa,Fc when scan rates were systemat-
ically made slower. For example, for 10, ipa,Rh/ipa,Fc
ratios (scan rates in mV sꢀ1 are given in brackets) were
found to be 0.77 (250), 0.75(200), 0.70(150), 0.63(100)
and 0.54(50), respectively (Fig. 5). This is a consequence
of more and more of an oxidised rhodium species being
generated by the equilibrium process, as scan rates
Fig. 6. Left. Coulometry experiments on [Rh(b-diketonato)(CO)2] solutions indicate that the overall amount of electrons that flows during oxidation
of 9 (d), 10 ( ) and 12 (m) is n = Q/NF = 3, 3 and 4, respectively. Inset. The definite biphasic current flow for the dfcm complex, 12, is consistent
¤
with two separate electron transfer processes taking place. The first is associated with the ferrocenyl groups, the second with RhI oxidation. Right.
CVs of samples taken from a solution of [Rh(FcCOCHCOCH3)(CO)2] on which bulk electrolysis was performed. Curve 1 (---) was obtained before
electrolyses commenced, i.e., at t = 0. Curve 2 (—) was obtained at t = 10 min, curve 3 (—) at t = 23 min after ca. 2 electrons per molecule has flown
and curve 4 (. . ..) at t = 90 min after 3 electrons has flown.

J. Conradie et al. / Inorganica Chimica Acta 358 (2005) 2530–2542
2537
decrease, at potentials close to Epa,Fc The result is deple-
tion of the remaining RhI species concentration and
hence the main electrochemical RhI oxidation wave
which is represented by the second anodic peak in the
CVs of 9–12 becomes relatively smaller. This observa-
tion must invariably point to the fact that ipc/ipa ratios
for the reversible ferrocenyl wave must come closer to
one at faster scan rates. This was indeed observed. Using
complex 10 again as an example, the ipc/ipa ratio, indi-
cated in brackets next to a specific scan rate (mV sꢀ1),
were as follows: 50 (0.83), 100 (0.85), 150 (0.88), 200
(0.90) and 250 (0.92).
To remove doubt that the ferrocene waves of 9–12 is
masking or hiding a small rhodium(I) oxidation wave
which overlaps with it, the CV of the ferrocene-free b-
diketonato complex [Rh(PhCOCHCOCF3)(CO)2] was
also recorded in acetonitrile (CVs not shown). Two irre-
versible RhI oxidation peaks were clearly observed: the
first at 0.727 and the second at 1.373 V versus Fc/Fc+.
When the CV of [Rh(PhCOCHCOCF3)(CO)2] was re-
corded in CH2Cl2, the two irreversible oxidation waves
collapsed into one broad peak at 1.191 V versus Fc/
Fc+. Complex 9, [Rh(FcCOCHCOCF3)(CO)2], only
showed rhodium(I) oxidation in dichloromethane at
1.195 V versus Fc/Fc+. These observations are consis-
tent with our findings in the cyclic voltammetry of the
present ferrocene-containing b-diketonato complexes
that the two rhodium oxidation processes are in equilib-
rium with each other in acetonitrile medium. A full elec-
trochemical study of ferrocene-free b-diketonato
dicarbonyl rhodium(I) complexes in coordinating and
non-coordinating solvents is at present underway. Fur-
ther proof that there is a second rhodium oxidation
wave overlapping with the ferrocenyl anodic wave will
be presented when CVs of the bulk electrolyses solution
at different time intervals during the course of the elec-
trolyses are discussed.
Having confirmed the slow dynamic equilibrium 1,
and the fact that their actually are two oxidation waves
for the RhI centre, one at potentials Epa,Rh > 0.7 V and
the other at potentials overlapping with the ferrocenyl-
based electrochemical waves, the observation that peak
2 becomes relatively stronger in relation to peak 1
(Fig. 2) at slower scan rates can now be explained. It
was already established that peak 1 is associated with
the redox activity of the ferrocenyl group of [Rh(FcC-
OCHCOCF3)(CO)2(CH3CN)]. The described electro-
chemical results is now consistent with the view that
the rhodium(I) core in the solvent-free species, [Rh(FcC-
OCHCOCF3)(CO)2], probably is oxidised at potentials
very close to that associated with peak 1, but before
peak 2 is invoked. This causes depletion of the equilib-
rium concentration of [Rh(FcCOCHCOCF3)(CO)2].
At fast scan rates, the equilibrium 1 is too slow to be
maintained and the observed intensity ratio for peaks
1 and 2 in Fig. 2 are indicative of the equilibrium
concentrations of [Rh(FcCOCHCOCF3)(CO)2] and
[Rh(FcCOCHCOCF3)(CO)2(CH3CN)]. However, at
slower scan rates, [Rh(FcCOCHCOCF3)(CO)2-(CH3CN)]
is noticeably dissociating to try and regenerate [Rh(FcC-
OCHCOCF3)(CO)2] according to equilibrium 1 as the
Rh centre of the solvent-free compound is electrochem-
ically oxidised. As additional [Rh(FcCOCHCOCF3)-
(CO)2] is formed from the dissociation process, its Rh
centre is continuously oxidised which leads to a build-
up of the rhodium-oxidised product, [Rhoxidised(FcC-
OCHCOCF3)(CO)2(solvent)x], before the potential is
reached to invoke formation of peak 2. Hence, when
the potential is reached to induce the oxidation of the
ferrocenyl group that is associated with [Rhoxidised(FcC-
OCHCOCF3)(CO)2 (solvent)x] at peak 2, a build-up of
the intermediate product [Rhoxidised(FcCOCHCOCF3)-
(CO)2 (solvent)x] has already taken place at slow scan
rates. This would explain why peak 2 draws relatively
more current under slow-scanning experimental condi-
tions. Arguments that the rhodium(I) core in [Rh(FcC-
OCHCOCF3)(CO)2(CH3CN)] is oxidised first, rather
than [Rh(FcCOCHCOCF3)(CO)2], would only be feasi-
ble if peak 1 is assigned to ferrocenyl oxidation of
[Rh(FcCOCHCO-CF3)(CO)2], but this would be in vio-
lation of the 1H NMR equilibrium study that was
performed.
The above described electrochemical results is consis-
tent with the following general electrochemical scheme
in the solvent acetonitrile:
[RhI(FcCOCHCOR)(CO)2(CH3CN)]
CH3CN + [RhI(FcCOCHCOR)(CO)2]
[RhI(Fc+COCHCOR)(CO)2(CH3CN)]
[Rhox(FcCOCHCOR)(CO)2(CH3CN)x]
[Rhox(Fc+COCHCOR)(CO)2(CH3CN)x]
To try and characterise the fully oxidised product in
the above scheme, samples was taken during different
time intervals of the bulk electrolyses experiments. CVs
and IR spectra were recorded of it, to try and establish
what the properties of the proposed final oxidised prod-
uct [Rhoxidised(Fc+COCHCOR)(CO)2 (solvent)x] are.
Fig. 6 (right) show these CVs for 10 at times t = 0, 10,
23 and 90 min of electrolyses. At 24 min 2 electrons were
counted, at 74 min and longer 3. Between 23 and 90 min
the bulk electrolyses solution changed colour via orange-
brown to brown black. It turns out that the electrochem-
ical generated product is unstable. It is clear that the
reduction peaks at Epc,Rh 6 ꢀ0.7 V becomes larger and
that the RhI oxidation waves at Epa,Rh P 0.7 V becomes
smaller during the initial stages of the electrolyses. After
23 min, the wave associated with Epa,Rh was for all prac-
tical purposes gone. The RhI oxidation peak that over-
lapped with the ferrocenyl peak was also disappearing
because ipa,Fc became smaller while ipc,Fc remained al-
most constant during the first 10-min period. Between
t = 10 and t = 23 min ipc,Fc did become slightly smaller,
but this is interpreted as evidence that the sample begins

2538
J. Conradie et al. / Inorganica Chimica Acta 358 (2005) 2530–2542
FcRhI ꢀ FcþRhI ꢀ FcþRhII; then
FcþRhII ꢀ FcRhIII !!! RhIII
to disintegrate. After 90 min of bulk electrolyses, the
ferrocenyl wave has also disappeared, indicating that
the entire molecule has fallen apart.
To determine whether the final rhodium oxidised spe-
cies still had carbonyl ligands bound to it, the IR study
of the bulk electrolyses solution was instructive. The IR
spectrum of bulk electrolyses solutions of 10 in acetoni-
trile in the presence of supporting electrolyte, but before
bulk electrolyses was initiated, showed two sharp CO
peaks at 2009 and 2083 cmꢀ1. In chloroform, in the ab-
sence of supporting electrolyte, these peaks are at 2014
and 2068, respectively. As bulk electrolyses continued,
under air free conditions and when one electron was
counted, weak peak pairs at 2015 and 1994 as well as
at 2087 and 2064 cmꢀ1 could be observed. When two
electrons was counted, the first-mentioned peak pair
was gone while the latter peak was low in intensity.
When three electrons were counted, no trace of any car-
bonyl peak was left on the IR. This indicates, as did the
CV, that the entire molecule has disintegrated, probably
to a RhIII species without any CO bound to it. The car-
bonyl IR peaks observed after electrolyses led to the
transfer of two electrons are not really in a position
where one would expect RhIII carbonyl peaks to be. It
has been observed before that CO stretching frequencies
shift about 76 cmꢀ1 to longer wave numbers upon mov-
ing from a RhI to a RhIII species [13]. This has been the
case in moving from [RhI(FcCOCHCOCH3)(CO)-
(PPh3)] with m(CO) = 1980 to [RhI(FcCOCHCOCH3)-
(CO)(PPh3)(CH3)(I)] with m(CO) = 2056 cmꢀ1. When
bulk electrolyses were performed but air (i.e., oxygen)
was allowed to enter the system, an IR set of peaks with
maxima at 2106 and 2137 were observed for a short time
during bulk electrolyses. However, they also disap-
peared completely by the time 3 electrons were counted
during electrolyses. A sample of the bulk electrolyses
solution after 2 electrons has flown was also taken and
tested for stability by IR monitoring. The complex(es)
in solution decomposed completely with no trace of
any CO peaks within 10 min.
Rhodium(II), a paramagnetic d7 species [24], can ex-
ist as a monomer, or as a dimer involving a Rh–Rh
bond. Neither IR nor anyone of the three electrochemi-
cal techniques utilised in this study clearly demonstrated
the existence of a rhodium(II) intermediate. Further
studies are required to determine the co-ordination
sphere and structure of the initially oxidised species be-
fore molecular disintegration takes place.
Finally, an attempt was made to correlate the ob-
served pK0a (Table 1) of the free b-diketones with the for-
mal reduction potentials of the ferrocenyl group of
complexes 9–12. As can be seen in Fig. 3, the formal
reduction potential of the ferrocenyl fragment is near-
linear and inversely proportional to pK0a of the free b-
diketones. This result implies that the group electroneg-
ativity of the R-groups, vR, and observed pK0a values of
b-diketones 1–4 are, as was found before [14], inversely
proportional to each other, and that this property is lar-
gely transferred to complexes of the type [Rh(FcCOCH-
COR)(CO)2]. However, the pK0a- and vR-potential
relationship breaks down when compared with the peak
anodic potentials of the Rh nucleus of these complexes
because of the irreversibility of the rhodium couples.
Since the sited rhodium anodic peak potentials does
not represent a true thermodynamic potential, little is
made of the observation that the anodic peak potential
of the rhodium centre does not correlate well with group
electronegativity or pKa values.
2.2. Crystal structure
To further the characterisation of complexes 9–12, we
present here the results of a single crystal structure
determination for 9. A perspective view showing atom
labelling of [Rh(FcCOCHCOCF3)(CO)2] is shown in
Fig. 7, crystal data, and selected bond lengths and angles
may be found in Tables 3 and 4, respectively. The struc-
ture showed a rotational as well as a translational dis-
placement disorder for the CF3 group. The rhodium
atom has a square planar co-ordination sphere. The
The above described IR study is consistent with a
RhII carbonyl species being formed initially before prod-
uct decomposition sets in to liberate a RhIII species with-
out any CO or ferrocene-containing ligands bound to it.
Rhodium(III) may be obtained directly in a two-electron
transfer process or it may be obtained after in situ gen-
erated rhodium(II) decomposes either in a dispropor-
tionation process, viz
˚
Rh(1)–O(3) bond is 0.033 A longer than the Rh–O(4)
bond, because of the stronger electron donating proper-
ties of the ferrocenyl group (vFc = 1.87) compared to
that of the CF3 group ðv
¼ 3:01Þ on the b-diketonato
CF3
ligand fctfa. One would also expect this electronic trans
influence of the ferrocenyl group to manifest in a shorter
Rh–C(12) bond, that is the bond trans to Rh–O(3) bond.
However, the poor accuracy of the bonds Rh–
FcRhI ꢀ FcþRhI ꢀ FcþRhII; then
2FcþRhII ꢀ FcþRhI þ FcþRhIII ! RhIII
˚
C(11) = 1.83(2) and Rh–C(12) = 1.84(1) A prevented
clear identification of this effect. An obvious feature of
the [Rh(FcCOCHCOCF3)(CO)2] complex was the inac-
curacy of many of the bond lengths. Bond lengths in
or via intramolecular electron transfer from the in situ
electrochemically generated rhodium(II) to the ferroce-
nium moiety

J. Conradie et al. / Inorganica Chimica Acta 358 (2005) 2530–2542
2539
Fig. 7. A perspective view of [Rh(fctfa)(CO)2] showing atom labelling. The view at right highlights the disordered trifluoromethyl group, the square
planar geometry around the Rh nucleus and the eclipsed conformation of the cyclopentadienyl groups.
Table 3
Crystal data for complex 9
Table 4
˚
Selected bond lengths (A) and angles (ꢀ) for complex 9
Empirical formula
Formula weight
Crystal size (mm3)
Crystal system
C16H10F3FeO4Rh
482.00
Bond lengths
Rh(1)–O(3)
Rh(1)–C(11)
O(1)–C(11)
O(3)–C(13)
C(6)–C(13)
C(14)–C(15)
2.049(8)
1.83(2)
1.14(1)
1.26(1)
1.46(2)
1.36(2)
Rh(1)–O(4)
Rh(1)–C(12)
O(2)–C(12)
O(4)–C(15)
C(13)–C(14)
2.016(9)
1.84(1)
1.12(2)
1.29(1)
1.42(2)
0.06·0.15·0.23
monoclinic
C2/c (15)
Space group
Unit cell dimensions
˚
a (A)
13.266(3)
19.553(3)
13.278(3)
100.92(2)
3382(1)
˚
b (A)
Bond angles
˚
c (A)
O(3)–Rh(1)–O(4)
O(3)–Rh(1)–C(12)
O(4)–Rh(1)–C(12)
Rh(1)–O(3)–C(13)
Rh(1)–C(11)–O(1)
C(1)–C(2)–C(3)
C(6)–C(13)–C(14)
C(7)–C(6)–C(13)
O(3)–C(13)–C(14)
90.1(4)
177.9(5)
89.9(5)
128.1(9)
179(1)
107(2)
120(1)
127(1)
124(1)
O(3)–Rh(1)–C(11)
O(4)–Rh(1)–C(11)
C(11)–Rh(1)–C(12)
Rh(1)–O(4)–C(15)
Rh(1)–C(12)–O(2)
C(7)–C(6)–C(10)
O(4)–C(15)–C(14)
C(10)–C(6)–C(13)
C(13)–C(14)–C(15)
91.0(5)
178.2(5)
89.0(6)
122.8(8)
179(1)
107(1)
131(1)
125(1)
125(1)
b (ꢀ)
3
˚
V (A )
Density (calculated) (g/cm3)
1.893
8
296(1)
Z
Temperature (K)
ꢀ1
˚
Mo Ka wavelength (A); l (cm
)
0.71069; 18.79
R = 0.057, Rw = 0.065
Final R indices [I > 3r(I)]a
P
P
a
R ¼ jjF oj ꢀ jF cjj= jF oj ¼ 0:0549 and
P
P
2
0:5
Rw ¼ ½ wðjF oj ꢀ jF cjÞ = wF 2oꢁ ¼ 0:0622.
nato)(CO)2] crystal structures are virtually unavailable.
The complexity of the disorder found in this particular
structure must also contribute to bond length and bond
angle inaccuracies. This inherent inaccurate Rh–C bond
lengths makes it impossible to forecast which carbonyl
group will be replaced by PPh3 in the substitution
reaction
complexes of the type [Rh(FcCOCHCOCF3)(PPh3)-
(CO)] [27], [Rh(FcCOCHCOCF3)(PPh3)(CO)(CH3)(I)]
[28], or [Rh(FcCOCHCOCF3)(cod)] [29], have been
determined more accurately. To our knowledge, the
only two other crystallographically characterised
complexes of the type [Rh(b-diketonato)(CO)2] are
[Rh(acetylacetonato)(CO)2] [30,31] and [Rh(benzoyltri-
fluoroacetonato)(CO)2] [32]. They showed similar large
errors as the complex in the present study in their bond
lengths and angles, see Table 5. The conclusion drawn is
that these large errors probably are due to an inherent
property of the sighted [Rh(b-diketonato)(CO)2] com-
plexes to form poor quality or weak crystals (we ob-
served significant crystal decay during data collection
for crystals) and probably explains why [Rh(b-diketo-
½Rhðb-diketonatoÞðCOÞ ꢁ þ PPh3
2
! ½Rhðb-diketonatoÞðPPh3ÞðCOÞꢁ þ CO
as a definite longer (i.e., weaker) Rh–C bond to the car-
bonyl group cannot be identified with certainty from
crystallographic data. Although the bond angles in the
rhodium polyhedron are within experimental error of
the expected ideal for dsp2 hybridisation (90ꢀ), devia-

2540
J. Conradie et al. / Inorganica Chimica Acta 358 (2005) 2530–2542
Table 5
Selected crystallographic data for [Rh(b-diketonato)(CO)2] complexes
0
0
Angle (C–Rh–C0) (ꢀ) Rh–O distance (A) Rh–O distance (A) Rh–C distance (A) Rh–C distance (A) Space group
˚
˚
˚
˚
Complex
ꢀ
[Rh(acac)(CO)2] [30] 85
[Rh(acac)(CO)2] [31] 88.9(3)
[Rh(tfba)(CO)2] [32] 87(1)
2.06
2.05
1.75
1.76
P1
ꢀ
2.044(4)
2.02(2)
2.049(8)
2.040(4)
2.02(2)
2.016(9)
1.831(7)
1.79(3)
1.84(1)
1.831(7)
1.82(3)
1.83(2)
P1
Pbac
C2/c
[Rh(fctfa)(CO)2]a
89.0(6)
O0 is the b-diketonato oxygen atom nearest to the most electronegative group on the b-diketonato ligand. C0 is the carbon atom of the carbonyl
group trans to O0.
a
This study.
tions of up to 11ꢀ were found in the sp2 hybridised C
atoms of the b-diketonato skeleton. The largest of these
is O(4)–C(15)–C(14) = 131(1)ꢀ. Although the free b-
diketonato ligand Hfctfa, 1 [14], and the cyclooctadiene
complex [Rh(FcCOCHCOCH3)(cod)], [29], were clearly
asymmetrically enolised or asymmetrically co-ordinated,
no asymmetric co-ordination pattern could clearly be
identified for the present complex, 9. The Rh. . ..Rh
3. Experimental section
3.1. Reagents
Literature methods were used to prepare 1–8 [11] and
9–12 [13]. Dry acetonitrile (Aldrich) was obtained by
refluxing under nitrogen over calcium hydride, distilled
onto alumina for storage, and redistilled just prior to use.
˚
distance is 3.346(2) A, which is incompatible with
metal–metal bonding. The C–C bond lengths of the
cyclopentadienyl rings (mean 1.41(3) with the longest
3.2. Electrochemistry
˚
and shortest bond lengths being 1.43(3) and 1.39(3) A,
respectively) are similar to the length of an aromatic
˚
C–C bond, 1.395(3) A [33]. The cyclopentadienyl rings
are almost planar with a dihedral angle of 2.34ꢀ and
for all practical purposes are in the eclipsed (D5h symme-
try) arrangement. The dihedral angle between the pseu-
do-aromatic system forming a plane defined by the
atoms C(13), C(14) and C(15), and the planar cyclopen-
tadienyl ring of the aromatic ferrocenyl moiety bound to
it, is 7.18ꢀ.
Cyclic voltammetry measurements on ca. 1 mmol
dmꢀ3 of solutions of 9–12 in dry acetonitrile containing
0.100 mol dmꢀ3 tetra-n-butylammonium hexafluoro-
phosphate ((nBu)4NPF6, Fluka, electrochemical grade)
as supporting electrolyte were conducted under a blan-
ket of purified argon at 25.0 ꢀC utilizing a BAS model
CV-27 voltammograph interfaced with a personal com-
puter. A three-electrode cell, which utilised a Pt auxil-
iary electrode, a glassy carbon working electrode, and
a Ag/Ag+ (0.0100 mol dmꢀ3 AgNO3) reference electrode
mounted on a Luggin capillary was used [14,34]. Data,
uncorrected for junction potentials, were collected with
an Adalab-PCꢂ and Adaptꢂ data acquisition kit (Inter-
active Microwave, Inc.) with locally developed software,
and analysed with Hyperplot (JHM International, Inc.).
In conclusion, it was demonstrated that ferrocenyl-
containing complexes of the type [Rh(FcCOCH-
COR)(CO)2] with R = CF3, CH3, Ph and Fc are oxidised
reversibly at less positive potentials than the rhodium
nucleus while a slow equilibrium process allows rho-
dium(I) oxidation at potentials associated with those
of ferrocene oxidation as well as at substantially more
positive potentials. Coulometric results in acetonitrile
were consistent with the oxidation of the rhodium(I)
centre to ultimately give a decomposed RhIII species
without any carbonyl or ferecene-containing ligand
bound to it. In acetonitrile solutions, complexes of the
type [Rh(FcCOCHCOR)(CO)2] are in slow equilibrium
with the adduct [Rh(FcCOCHCOR)(CO)2(CH3CN)].
The latter is the dominant species at lower concentra-
tions. Electronic communication between the Fc frag-
ment and the R-groups via the pseudo-aromatic b-
diketonato core of the complexes were efficient enough
to allow a linear relationship between group electroneg-
All cited potentials were referenced against the Fc/Fc+
+
couple, which exhibited E0 = 0.087 V versus Ag/Ag ,
0
ipc/ipa = 0.98, DEp = 74 mV under our experimental con-
ditions. OSWV measurements were perform under the
same conditions as described for CV measurements
utilising a BAS 100B/W electrochemical workstation
interfaced with a personal computer.
Bulk electrolyses were carried out utilising a BAS
CV-27 voltammograph at 25.0(1) ꢀC in acetonitrile. A
three-electrode cell equipped with a Pt wire auxiliary
electrode (isolated from the sample by means of a
0.1 mol dmꢀ3 (nBu)4NPF6/CH3C N salt bridge), a glassy
carbon working electrode (electro-active area 3 cm2)
and a Ag/Ag+ (0.010 mol dmꢀ3 AgNO3 in 0.1 mol
dmꢀ3 (nBu)4NPF6/CH3CN) reference electrode mounted
on a Luggin capillary were employed. Current readings
and the integrated current (coulombs) were recorded
manually at different times, t, during the course of the
experiments.
ativity, vR, and E0 of the ferrocenyl group and an al-
0
most linear relationship between free b-diketone pKa
and E0 of the ferrocenyl group. The relationship be-
0
tween anodic potential potentials of the Rh nuclei and
vR as well as b-diketone pKa are non-specific.

J. Conradie et al. / Inorganica Chimica Acta 358 (2005) 2530–2542
2541
[3] G.J.J. Steyn, A. Roodt, J.G. Leipoldt, Inorg. Chem. 31 (1992)
3477.
3.3. X-ray structure determination
[4] J.G. Leipoldt, S.S. Basson, J.J.J. Schlebusch, E.C. Grobler, Inorg.
Chim. Acta 62 (1982) 113.
Crystals of 9 were obtained from hot hexane by slowly
cooling the reaction mixture. A dark red block crystal was
mounted on a glass fiber rod and intensity data were col-
lected on a Rigaku AFC5R diffractometer using graphite
monochromated Mo Ka radiation and a rotating anode
generator. Cell constants and an orientation matrix for
data collection, obtained from a least-squares refinement
using the setting angles of 25 carefully centred reflections
in the range 28.44ꢀ < 2h < 36.52ꢀ, corresponded to a C-
centred monoclinic cell. Of the 5291 reflections which
were collected, 5091 were unique (Rint = 0.040). The lin-
ear absorption coefficient, l, for Mo Ka radiation was
18.79 cmꢀ1. An empirical absorption correction based
on azimuthal scans of several reflections was applied
which resulted in transmission factors ranging from 0.69
to 1.00. The data were corrected for Lorentz and polarisa-
tion effects.
The structure was solved by direct methods (SHELXS-
86) [35] and expanded using Fourier techniques (DIRDIF-
94) [36]. Some non-hydrogen atoms were refined aniso-
tropically, while the rest were refined isotropically.
Hydrogen atoms were included but not refined. The
CF3 group is disordered with both rotational disorder
and with the carbon atom in slightly different positions.
The groups were refined as two rigid groups with isotropic
thermal parameters and with the C–F distances defined as
[5] C.E. Hickey, P.M. Maitlis, J. Chem. Soc., Chem. Commun.
(1984) 1609.
[6] A. Borrini, G. Ingrosso, J. Organomet. Chem. (1977) 275.
[7] S. Back, R.A. Gossage, H. Lang, G. van Koten, Eur. J. Inorg.
Chem. (2000) 1457.
[8] M.I. Bruce, B.C. Hall, P.J. Low, M.E. Smith, B.W. Skelton,
A.H. White, Inorg. Chim. Acta 300 (2000) 633.
[9] M. Laly, R. Broussier, B. Gautheron, Tetrahedron Lett. 41 (2000)
183.
[10] L. Hao, M.A. Mansour, R.J. Lachicotte, H.J. Gysling, R.
Eisenberg, Inorg. Chem. 39 (2000) 5520.
[11] W.C. Du Plessis, T.G. Vosloo, J.C. Swarts, J. Chem. Soc., Dalton
Trans. (1998) 2507.
[12] T.G. Vosloo, W.C. Du Plessis, J.C. Swarts, Inorg. Chim. Acta
331 (2002) 188.
[13] J. Conradie, G.J. Lamprecht, S. Otto, J.C. Swarts, Inorg. Chim.
Acta 328 (2002) 191.
[14] W.C. Du Plessis, J.J.C. Erasmus, G.J. Lamprecht, J. Conradie,
T.S. Cameron, M.A.S. Aquino, J.C. Swarts, Can. J. Chem. 77
(1999) 378.
[15] See for example C. Creutz, H. Taube, J. Am. Chem. Soc. 91
(1969) 3988.
[16] W.E. Geiger, N. Van Order, D.T. Pierce, T.E. Bitterwolf,
A.L. Reingold, N.D. Chasteen, Organometallics 10 (1991)
2403.
[17] N. Van Order, W.E. Geiger, T.E. Bitterwolf, A.L. Reingold,
J. Am. Chem. Soc. 109 (1987) 5680.
[18] D.T. Pierce, W.E. Geiger, Inorg. Chem. 33 (1994) 373.
[19] T.G. Vosloo, J.C. Swarts, Trans. Met. Chem. 27 (2002)
411.
˚
1.31 A. The final cycle of full-matrix least-squares refine-
[20] See for example V.F. Kuznetsov, G.A. Facey, G.P.A. Yap, H.
Alper, Organometallics 18 (1999) 4706.
ment on F was based on 1418 observed reflections
(I > 3.00r(I)) and 194 variable parameters. All calcula-
tions were performed using the teXsan [37] crystallo-
graphic software package of the Molecular Structure
Corporation. Relevant crystal data are given in Table 3.
[21] S.S. Basson, J.G. Leipoldt, J.A. Venter, Acta Crystallogr. C 46
(1990) 1324.
[22] P.W. Atkins, Physical Chemistry, fifth ed., Oxford University
Press, Oxford, 1994, p. 288.
[23] P.M. Bush, J.P. Whitehead, C.C. Pink, E.C. Gramm, J.L. Eglin,
S.P. Watton, L.E. Pence, Inorg. Chem. 40 (2001) 1871.
[24] F.A. Cotton, G. Wilkinson, C.A. Murillo, M. Bochman,
Advanced Inorganic Chemistry, sixth ed., John Wiley and Sons,
New York, 1999, p. 1053.
Acknowledgements
[25] V. Gutman, The Donor–acceptor Approach to Molecular Inter-
actions, Plenum Press, New York, 1978, p. 20.
[26] F.A. Cotton, G. Wilkinson, C.A. Murillo, M. Bochman,
Advanced Inorganic Chemistry, sixth ed., John Wiley and Sons,
New York, 1999, p. 1053.
The authors gratefully acknowledge financial assis-
tance from the Foundation for Research and Develop-
ment and the Central Research Fund of the University
of the Free State.
[27] G.J. Lamprecht, J.C. Swarts, J. Conradie, J.G. Leipoldt, Acta
Crystallogr. C 49 (1993) 82.
[28] J. Conradie, A. Roodt, G.J. Lamprecht, J.C. Swarts, unpublished
results.
Appendix A. Supplementary data
[29] J.C. Swarts, T.G. Vosloo, J.G. Leipoldt, G.J. Lamprecht, Acta
Crystallogr. C 49 (1993) 760.
Supplementary data associated with this article can
be found, in the online version, at doi:10.1016/
j.ica.2005.02.010.
[30] F. Huq, A.C. Skapski, J. Cryst. Mol. Struct. 4 (1974) 411.
[31] N.A. Bailey, E. Coates, G.B. Robertson, F. Bonati, R. Ugo,
Chem. Commun. (1967) 1041.
[32] J.G. Leipoldt, L.D.C. Bok, S.S. Basson, J.S. Van Vol-
lenhoven, T.I.A. Gerber, Inorg. Chim. Acta 25 (1977) L63–
L64.
References
[33] R.C. Weast, Handbook of Chemistry and Physics, 63rd
ed., The Chemical Rubber Company, Cleveland, OH, p.
F180.
[1] J. Jones, Platinum Met. Rev. 44 (2000) 94.
[2] J.A. Venter, J.G. Leipoldt, R. van Eldik, Inorg. Chem. 30 (1991)
2207.
[34] J.C. Swarts, E.H.G. Langner, N. Krokeide-Hove, M.J. Cook, J.
Mater. Chem. 11 (2001) 434.

2542
J. Conradie et al. / Inorganica Chimica Acta 358 (2005) 2530–2542
[35] SHELXS-86: G.M. Sheldrick, in: G.M. Sheldrick, C. Kruger, R.
Goddard (Eds.), Crystallographic Computing 3, Oxford Univer-
sity Press, Oxford, 1985, pp. 175–189.
Program System, Technical report of the Crystallography Labo-
ratory, University of Nijmegen, Nijmegen, The Netherlands, 1994.
[37] TeXan for windows version 1.05: Crystal structure analyses
package, Molecular Structure Corporation, The Woodlands, TX,
1997.
[36] DIRDIF-94: P.J. Beurskens, G. Admiraal, G. Beurskens, W.P.
Bosman, R. de Gielder, R. Israel, J.M.M. Smits, The DIRDIF-94