Chemical ligand non-innocence in pyridine diimine Rh complexes

Article abstract of DOI:10.1016/j.ica.2004.02.012

The formation and reactivity of pyridine diimine rhodium(I) alkyl complexes without β-hydrogens (Me, Bz, CH₂SiMe₃) is described. In contrast to the corresponding cobalt complexes, the rhodium complexes could not be activated to polymerise ethene. Rh ethyl complexes could not be prepared. Examples of hydrogen transfer to and from the ligand were observed, illustrating the active role the pyridine diimine ligand can play in the reactions of its complexes. Decomposition via loss of free ligand was observed in many cases, indicating that the pyridine diimine ligand is not a very suitable one for Rh^I.

Full text of DOI:10.1016/j.ica.2004.02.012

Inorganica Chimica Acta 357 (2004) 2945–2952
www.elsevier.com/locate/ica
Chemical ligand non-innocence in pyridine diimine Rh complexes
T. Martijn Kooistra, Dennis G.H. Hetterscheid, Erik Schwartz, Quinten Knijnenburg,
*
Peter H.M. Budzelaar , Anton W. Gal
Metal-Organic Chemistry, University of Nijmegen, Toernooiveld 1, 6525 ED Nijmegen, The Netherlands
Received 16 January 2004; accepted 14 February 2004
Available online 12 March 2004
Abstract
The formation and reactivity of pyridine diimine rhodium(I) alkyl complexes without b-hydrogens (Me, Bz, CH2SiMe3) is de-
scribed. In contrast to the corresponding cobalt complexes, the rhodium complexes could not be activated to polymerise ethene. Rh
ethyl complexes could not be prepared. Examples of hydrogen transfer to and from the ligand were observed, illustrating the active
role the pyridine diimine ligand can play in the reactions of its complexes. Decomposition via loss of free ligand was observed in
many cases, indicating that the pyridine diimine ligand is not a very suitable one for RhI.
Ó 2004 Elsevier B.V. All rights reserved.
Keywords: Ligand non-innocence; Rhodium; Pyridine diimine; Alkyl
1. Introduction
Electronic non-innocence of the pyridine diimine has
been demonstrated in several cases [6] and is by now an
established pattern in the chemistry of this ligand. The
relevance of this form of non-innocence to catalysis is
still unclear, but its stabilising role on the way to active
catalysts may well be important.
Another emerging theme in the chemistry of pyridine
diimine complexes is the involvement of the ligand in
chemical transformations. Whereas with more tradi-
tional ligands (cyclopentadienyls, phosphines) reactions
normally happen at the metal, complexes of the pyridine
diimine ligand frequently undergo attack at the ligand,
leading to alkylation [7] (at the imine carbon or the
pyridine 1, 2 or 4 position) or deprotonation [8]; sub-
sequent dimerisation of the ligand skeleton has also been
observed [8b]. Fig. 1 summarises a few of the more
unusual examples.
From these, it is clear that the ligand can play a role
that goes beyond the normal spectator role and can re-
ally be called ‘‘chemically non-innocent’’. One could
argue that electronic and chemical non-innocence are
related, in the sense that unpaired electron density on
the ligand would increase its reactivity. If that is the
case, the metal should also be important, since it de-
termines the amount of electron density transferred to
the ligand and the amount of radical character involved
For a long time, olefin polymerisation catalysis was
believed to be restricted to early transition metals. The
situation changed with the discovery of fairly efficient Ni
and Pd catalysts bearing bulky diimine ligands [1]. Even
more spectacular was the simultaneous discovery by
BrookharT and Gibson [2] of Fe and Co catalysts bearing
pyridine diimine (N3) ligands. Whereas the catalysis by
Ni/Pd catalysts appears to be straightforward and is un-
derstood in considerable detail, the Fe and Co systems are
still shrouded in mystery. The active species remains un-
known for both metals. For Fe, one generally assumes a
cationic alkyl complex LFeRþ to be involved, but direct
evidence is lacking and FeIII has also been suggested as
active species [3]. Even if the active species is indeed
LFeRþ, there is uncertainty about its spin state [4].
For Co, the situation is also unclear, but parts of the
activation route have been identified. The ligand plays a
prominent role here in stabilising the unusual square-
planar CoI precursors [5]. This happens through ac-
cepting an electron from the metal, i.e., the ligand is
‘‘non-innocent’’ in an electronic sense.
* Corresponding author. Tel.: +31-24-3652186; fax: +31-24-3553450.
E-mail address: budz@sci.kun.nl (P.H.M. Budzelaar).
0020-1693/$ - see front matter Ó 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.ica.2004.02.012

2946
T.M. Kooistra et al. / Inorganica Chimica Acta 357 (2004) 2945–2952
R
N
R
N
N
N
N
Me
VI
Ar
MeLi
Ar
N
N
N
N
Al
Me Me
Me
Ar
Ar
Ar
Ar
Me
Li
OEt2
Li(THF)3
-
Me3Al
(a)
Me
Me
∆
1) VCl3
2) MAO/
MeLi
Me
N
N
N
MeLi
Ar
N
N
N
N
Me
Li
VIII
Cl Cl
VI
Me Me
(b)
MeLi
(e)
Ar
Ar
Ar
N
N
N
Ar
Ar
N
N
Ar
Ar
OEt2
1) MnCl2
2) RLi
1) CrCl3
2) BzMgCl
(d)
Ar
(c)
Bz
Ar
N
N
_
N
N
N
N
N
N
MnI
MnI
R
Ar
Ar Ar
Ar
Bz
Cr
N
R
N
N
N
Cr
Bz
N
N
MnII
R
N
Ar
Ar
Ar
Ar
Bz
Fig. 1. Examples of ligand-centred reactivity of pyridine diimine complexes. (a) [7c]; (b) [7d]; (c) [8]; (d) [8b]; (e) [7a,7b].
[6b], both of which should influence ligand centred re-
activity. The variety in reactivity and reactive sites in
Fig. 1 indicates that this indeed is the case.
In the present paper, we describe a study of RhI alkyl
derivatives of the pyridine diimine ligand.
NaOMe
Ar
1/3 Me3Al
Ar Ar
N
N
N
N
N
N
N
N
N
Rh
Cl
Rh
Ar
Ar
Rh
Me
Ar
OMe
Ar =
Following our investigations of the Co polymerisa-
tion system [5b], we decided to study the corresponding
Rh complexes, motivated by the formal analogy be-
tween square-planar LRhX and LCoX species. Several
pyridine diimine RhI complexes have been reported [9]
and LRhCl is easily synthesised from the ligand and
[Rh(C2H4)2Cl]2 (Fig. 2).
Dias and his coworkers [9b] used these complexes as
the starting point for studies aimed at ethene polymer-
isation. They were able to synthesise cationic RhIII al-
kyl-olefin complexes by a series of substitution and
oxidative-addition reactions, but were unable to induce
polymerisation. We believe that the corresponding RhI
alkyls would be better models for relevant Co species
than the RhIII alkyls studied by Dias. In any case, a
comparison of the chemical behaviour of the corre-
sponding CoI and RhI alkyls (e.g., in their reactivity
towards olefins) should help us understand to what ex-
Fig. 3. Synthesis of L0RhMe [10].
tent the diimine pyridine ligand can help making Co
behave like a second-row transition metal in adapting a
square-planar geometry and a formal +1 oxidation
state.
Burger and his coworkers already published the
synthesis of L0RhMe containing 2,6-dimethylphenyl
groups at N. This complex could not be obtained via
direct alkylation of the chloride derivative [10]; instead,
they first synthesised the methoxy complex and con-
verted this to the methyl complex (Fig. 3). The corre-
sponding iridium complex L0IrMe was synthesised in the
same way.
2. Experimental
R2
N
R2
All manipulations were performed under an inert
atmosphere using standard Schlenk techniques or in a
dry-box. Solvents were purified by distillation from an
appropriate drying agent. All NMR spectra were re-
corded in d6-benzene at room temperature on Bruker
200, 300 and 500 MHz or Varian Inova 400 MHz
spectrometers. [Rh(C2H4)2Cl]2 [11], 1a [9b], L2 [10] were
synthesised according to the literature procedures.
[Rh(C2H4)2Cl]2
- C2H4
N
N
N
N
N
Rh
Cl
Ar
Ar
Ar
Ar
L1: R2 = H
L2: R2 = tBu
1a,b
Ar =
Fig. 2. Synthesis of LRhCl complexes.

T.M. Kooistra et al. / Inorganica Chimica Acta 357 (2004) 2945–2952
2947
2.1. L1RhMe (2a)
140.6 (Ar o), 126.7 (Ar p), 124.3 (Ar m), 123.6 (Py 3),
3
28.2 (CHMe2), 24.5, 24.3 (CHMe2), 19.9 (d, JRhC 2.0
1
Two millilitres of toluene was added to 100 mg of 1a
(0.16 mmol) and 7.1 mg MeLi (0.32 mmol; 2 equiv.) and
the mixture was stirred for 3 days. The solution was
filtered and the solvent was removed in vacuo, yielding
91 mg of 2a as a dark green solid.
Hz, N@CMe), 6.7 (d, JRhC 33 Hz, RhCH2SiMe3), 4.0
(RhCH2SiMe3).
2.4. L1RhX (5a)
1H NMR (300 MHz): d 8.19 (t, JHH 7.9 Hz, 1H, Py
In an NMR tube equipped with a septum, a few
milligrams of 2a, 3a or 4a were dissolved in 600 ll C6D6.
Hydrogen (1 ml) was injected in the NMR tube and a
colour change from dark green to red-brown was ob-
served (yellow green when traces of 1a were present in
the sample).
3
4), 7.34 (d, 3JHH 7.7 Hz, 2H, Py 3), 7.22–7.10 (m, 6H, Ar
3
m,p), 3.19 (sept, JHH 6.8 Hz, 4H, CHMe2), 2.12 (d,
2JRhH 1.1 Hz, 3H, RhCH3), 0.99 (m, 24H, CHMe2), 0.73
(s, 6H, N@CMe).
13C NMR (75 MHz): d 166.8 (C@N), 155.3 (d, JRhC
2
2.7 Hz, Py 2), 148.3 (Ar i), 140.6 (Py 4), 140.4 (Ar o),
123.6 (Ar m), 123.1 (Py 3), 28.2 (CHMe2), 23.9, 23.6
(CHMe2), 18.9 (N@CMe), 1.2 (d, 1JRhC 21 Hz, RhCH3).
2.5. (L1 + 4H)RhX (6a)
Two millilitres of toluene and 0.3 ml of 1.1 M Et2Zn
in toluene (0.33 mmol; 4 equiv.) were added to 100 mg
1a (0.16 mmol). The mixture was stirred for 24 h and
subsequently filtered. The solvent was removed in vacuo
and the residue was dissolved in C6D6. NMR showed a
mixture 6a and free ligand. After a few days only free
ligand remained.
2.2. L1RhBz (3a)
Two millilitres of toluene was added to 100 mg of 1a
(0.27 mmol) and 41 mg of Bz2Mg (0.27 mmol; 2 equiv.).
The mixture was stirred for 24 h, filtered and the filtrate
dried in vacuo. The resulting purple powder was ex-
tracted with hexane and the solvent was removed in
vacuo, giving 3a as a purple solid.
1H NMR (200 MHz): d 7.31–6.93 (m, 6H, Ar m,p),
5.09 (t, 3JHH 4.0 Hz, 2H, Py0 3), 3.56 (t, 3JHH 4.0 Hz, 2H,
Py0 4), 2.73 (sept, 3JHH 5.1 Hz, 4H, CHMe2), 1.67 (s, 6H,
1H NMR (200 MHz): d 7.97 (t, JHH 7.9 Hz, 1H, Py
3
3
4), 7.31–7.17 (m, 8H, Ar m,p and Py 3), 7.21 (t, 3JHH 7.3
N@CMe), 1.20, 1.00 (d, JHH 6.9 Hz, 12H each,
CHMe2).
Hz, 2H, Bz m), 6.65 (m, 1H, Bz p), 5.59 (d, 3JHH 6.9 Hz,
2
2H, Bz o), 3.69 (d, JRhH 2.6 Hz, 2H, RhCH2C6H5),
3
3.28–3.05 (sept, JHH 8.0 Hz, 4H, CHMe2), 1.01–0.92
2.6. (L1 ) H)Rh(C2H4) (7a)
(m, 30H, N@CMe and CHMe2).
13C NMR (75 MHz): d 166.8 (d, 2JRhC 2.3 Hz, C@N),
2.6.1. With diethylzinc
2
156.2 (d, 2JRhC 2.9 Hz, Py 2), 151.9 (d, JRhC 1.7 Hz, Bz
Two millilitres of cyclohexane and 0.3 ml of 1.1 M
Et2Zn in toluene (0.33 mmol; 4 equiv.) were added to
100 mg 1a (0.16 mmol). The mixture was stirred for 24 h
and subsequently filtered. NMR showed a mixture 7a
and free ligand. After a few days only free ligand
remained.
i), 148.4 (Ar i), 141.0 (Py 4), 140.6 (Ar o), 127.3 (Bz o),
127.0 (Bz m), 126.5 (Ar p), 124.2 (Ar m), 123.6 (Py 3),
119.6 (Bz p), 28.6 (CHMe2), 24.0, 23.8 (CHMe2), 19.5
(d, 3JRhC 1.7 Hz, N@CMe). The benzylic CH2 signal was
not observed.
2.3. L1RhCH2SiMe3 (4a)
2.6.2. With hydrogen/ethene
An excess of ethene was added with a syringe to a
sample of 5a in an NMR tube equipped with a septum.
The NMR tube was placed under an argon atmosphere
for several hours and subsequently an NMR spectrum
was recorded.
Four millilitres of toluene and 0.54 ml of 1.0 M
Me3SiCH2Li in pentane (0.54 mmol; 2 equiv.) were
added to 170 mg of 1a (0.27 mmol). The mixture was
stirred for 24 h, filtered and the filtrate was dried in
vacuo, yielding a green-brown powder. Extraction with
hexane and removal of the solvent resulted in a quan-
titative yield of 4a as a green powder.
1H NMR (400 MHz): d 7.3–7.0 (br, 9H, Ar m,p and
Py 3,4,5), 4.56 (s, 1H, trans CH@C–N), 4.12 (s, 1H, cis
CH@C–N), 3.80 (d, 4H, 2JRhH 1.8 Hz, C2H4), 3.63, 2.94
3
3
1H NMR (300 MHz): d 8.06 (t, JHH 7.9 Hz, 1H, Py
(sept, JHH 7.0 Hz, 2H each, CHMe2), 2.27 (s, 3H,
3
4), 7.24 (d, 3JHH 8.1 Hz, 2H, Py 3), 7.15–7.03 (m, 6H, Ar
N@CMe), 1.34, 1.27, 1.24, 0.81 (d, JHH 6.8 Hz, 6H
each, CHMe2).
3
m,p), 3.26 (sept, JHH 6.8 Hz, 4H, CHMe2), 1.87 (d,
3
2JRhH 2.2 Hz, 2H, RhCH2), 1.28, 0.97 (d, JHH 6.6 Hz,
12H each, CHMe2), 0.86 (s, 6H, N@CMe), 0.29 (s, 9H,
SiMe3).
2.7. L2RhCl (1b)
13C NMR (75 MHz): d 165.8 (d, 2JRhC 0.9 Hz, C@N),
A mixture of 1.1 g (2.0 mmol; 1 equiv.) L2 and 0.4 g
(1.0 mmol) [Rh(C2H4)2Cl]2 in 12 ml toluene was stirred
2
156.3 (d, JRhC 3.2 Hz, Py 2), 149.4 (Ar i), 141.1 (Py 4),

2948
T.M. Kooistra et al. / Inorganica Chimica Acta 357 (2004) 2945–2952
3
for 30 min. Removal of the solvent in vacuo gave 1b as a
dark green solid, which was used without further puri-
fication. The yield was 1.2 g (1.7 mmol; 83%).
0.84 (d, JHH 6.8 Hz, 6H each, CHMe2), 1.11 (s, 9H,
CMe3), 0.83 (s, 3H, N–CMe).
1H NMR (300 MHz): d 7.29 (s, 2H, Py 3), 7.15–6.95
3
(m, 6H, Ar m,p), 3.24 (sept, JHH 6.9 Hz, 4H, CHMe2),
3. Results
3
1.36, 1.07 (d, JHH 6.9 Hz, 12H each, CHMe2), 1.10 (s,
9H, CMe3), 1.10 (s, 6H, N@CMe).
3.1. Basic ligand system L1
13C NMR (75 MHz): d 166.4 (d, 2JRhC 2.3 Hz, C@N),
4
156.4 (d, JRhC 3.2 Hz, Py 4), 146.6 (Py 2), 146.3 (Ar i),
140.5 (Ar o), 127.0 (Ar p), 123.6 (Ar m), 121.1 (Py 3),
3.1.1. Synthesis of rhodium alkyl complexes
L1RhCl (1a) was easily synthesised according to the
literature procedures [9b]. In contrast to L0RhCl [10], 1a
could easily be alkylated using organolithium or or-
ganomagnesium compounds. L1RhMe (2a), L1RhBz
(3a) and L1RhCH2SiMe3 (4a) were obtained in good
yields from the reaction of 1a with MeLi, Bz2Mg and
LiCH2SiMe3, respectively. The two most likely side re-
actions in these alkylations are addition to the imine
functionality or deprotonation of the ketimine methyl
group. It seems probable that either or both of these
caused the problems encountered by Burger. The larger
isopropyl group in the ligand we used may cause just
enough steric protection of the imine to prevent these
reactions. When a large excess of alkylating agent was
used, we obtained mixtures of various unidentified
products, possibly formed via attack on the imine and/
or deprotonation of the ligand. We were not able to
crystallise any of the alkyl complexes due to their in-
stability in solution. In hydrocarbons, a film or deposit
of metallic rhodium was usually formed on standing. In
THF the complexes reacted to a mixture of various
products, while in dichloromethane 1a was formed in-
stantly from the methyl and benzyl complexes. Proba-
bly, dichloromethane oxidatively adds to rhodium and
subsequently a chlorinated alkane is eliminated, as was
reported before for related complexes [9a]. In contrast
with the report by Dias [9b], even CD2Cl2 was found to
react with 1a, albeit very slowly: after one hour, 10%
37.1 (CMe3), 29.7 (CMe3), 29.0 (CHMe2), 24.4, 24.3
3
(CHMe2), 17.8 (d, JRhC 2.1 Hz, N@CMe).
2.8. L2RhMe (2b)
Ten millilitres of toluene was added to 108 mg of 1b
(0.15 mmol) and 6.8 mg of MeLi (0.31 mmol; 2 equiv.).
The mixture was stirred for 7 days, filtered and the fil-
trate dried in vacuo. The resulting dark green solid was
used without further purification.
1H NMR (200 MHz): d 7.77 (s, 2H, Py 3), 7.25–7.10
3
(m, 6H, Ar m,p), 3.26 (sept, JHH 6.8 Hz, 4H, CHMe2),
2.12 (d, 2JRhH 1.2 Hz, 3H, RhCH3), 1.24 (s, 9H, CMe3),
3
1.14, 1.08 (d, JHH 6.8 Hz, 12H each, CHMe2), 0.86 (s,
6H, N@CMe).
2.9. (L2 ) H)Rh(C2H4) (7b)
2.9.1. With diethylzinc
Ten millilitres of toluene was added to 100 mg of 1b
(0.15 mmol) and 63 mg of Et2Zn (0.44 mmol; 6 equiv.).
The dark yellow/green mixture was stirred for 12 h, fil-
1
tered and the filtrate dried in vacuo. H NMR (C6D6)
showed the presence of 7b as well as several other
products.
2.9.2. With hydrogen/ethene
In an NMR tube, ca. 5 mg of 2b was dissolved in
C6D6. Hydrogen (1 ml) was injected and a colour
change from green to brown/green was observed. Next
an excess of ethene was added through the septum, and
was converted to
a
new complex, possibly
L1RhCl2(CD2Cl). Oxidative addition of C–Cl bonds to
rhodium(I) chloride complexes is not unusual [9a].
1
subsequently an H NMR spectrum was recorded.
3.1.2. Reactivity of rhodium alkyl complexes
Because both L1CoCl and L1CoR polymerise ethene
after activation with MAO, we tested whether the cor-
responding Rh complexes could be activated in the same
way. After treatment with MAO under 7 bar of ethene,
no polymer was formed. Because of the similarity of the
complexes, this complete absence of activity was not
expected.
Pyridine diimine CoI alkyl complexes react with hy-
drogen to LCoH [12]. This hydride complex is an active
catalyst for the hydrogenation of mono- and di-substi-
tuted olefins, but not for tri-substituted olefins. When
L1RhR complexes 2a–4a were reacted with H2, the
NMR spectrum did not show any signals for the
expected diamagnetic hydride complex. Instead, only
2.9.3. With LDA/ethene
To a solution of 100 mg (0.15 mmol) 1b in toluene
was added an equimolar amount of an LDA solution in
THF (2.5 ml) at )15 °C. The mixture was stirred vig-
orously under an ethene atmosphere for several minutes.
The red solution was warmed to room temperature and
stirred for another 20 min in which it turned yellow.
Subsequently, the solvent was removed in vacuo.
1H NMR (400 MHz): d 8.11 (d, JHH 1.2 Hz, 1H, Py
4
3), 7.66 (d, 4JHH 1.2 Hz, 1H, Py 5), 7.20–7.05 (m, 6H, Ar
m,p), 4.67 (s, 1H, trans CH@C–N), 4.13 (s, 1H, cis
CH@C–N), 3.77 (d, JRhH 2.0 Hz, 4H, C2H4), 3.71, 3.02
3
(sept, JHH 6.8 Hz, 2H each, CHMe2), 1.38, 1.30, 1.27,

T.M. Kooistra et al. / Inorganica Chimica Acta 357 (2004) 2945–2952
2949
signals for impurities in the starting material and a very
small amount of asymmetric product were visible, to-
gether with signals for RH. Most of the alkyl complex
was obviously transformed into an NMR-silent para-
magnetic species. This was confirmed by the presence of
a very intense signal in the EPR spectrum of the product
(5a) (Fig. 4).
hydrogen, the same selective hydrogenation was ob-
served as for L1CoH [12] that is, mono- and di-substi-
tuted olefins were hydrogenated, but tri-substituted
olefins were not. This lead us to conclude that complex
5a, rather than Rh0 formed by its slow decomposition, is
responsible for the activity.
The nature of complex 5a is not clear. It most likely
has the composition L1RhX, where X is arene (solvent),
N2, H2 or two hydrides. Burger has observed formation
of L0Rh(l-N2)RhL0 complexes [10b], but a binuclear
structure would not be expected to give the type of EPR
signal we observe. Assuming for the moment a compo-
sition LRhX with a neutral ligand X, the g-values of
2.017 and 1.978 suggest that the unpaired electron is
located partly on the organic ligand and partly on the
rhodium centre. This indicates that the bonding is in-
termediate between a true Rhð0Þ complex L1Rhð0Þ and a
complex of RhI with a ligand radical anion, i.e.
L1ðÅꢀÞRhI. Formation of the L1ðÅꢀÞ radical anion has
been observed before, e.g., in complexes with low-valent
Co and Mn [5,8]. Solutions of 5a slowly deposit a film of
metallic rhodium.
A possible mechanism for the formation of 5a is
outlined in Fig. 5. The combination of oxidative addi-
tion and comproportionation leads to the formation of
two molecules of a RhII complex, which eliminates
methane to give the Rh0 final product. It is not obvious
how the comproportionation step occurs, since the li-
gands are very large and should shield the metal centre
enough to hinder the close approach required for direct
hydrogen transfer. It may be that a hydrogen atom is
transferred via the pyridine rings of the ligands (see
below).
3.1.3. Attempted synthesis of L1RhEt
L1CoEt is easily formed from L1CoH and ethene [12].
If the reaction of 1a with Et2Zn was carried out in
benzene or toluene, initially broad lines were observed in
1
the H NMR spectrum, but after a few hours a dia-
magnetic product (6a) had formed; a small amount of
ethene could also be seen. Product 6a showed two
triplets at 3.6 and 5.1 ppm (2H each). We believe that
the only reasonable explanation is that these signals
belong to
a 1,4-dihydro-1-pyridyl group, formed
through migration of a hydrogen atom from the metal
1
to the 4-position of the pyridine ring. The distances
involved make direct intramolecular transfer of hydro-
gen to the 4-position unlikely; the hydrogen may have
moved from the metal either in a stepwise fashion (i.e.,
via some of the other ring carbons) or via intermolecular
hydrogen transfer.
Presumably, in 6a an additional ligand (e.g., solvent
or ethene) also coordinates to the metal (see Fig. 6). It
was not possible to identify this ligand from the 1H
2
NMR spectrum. Full characterisation of 6a was not
possible, because on standing in solution it slowly de-
composed to the free ligand and metallic rhodium.
The most likely path to 6a is initial formation of
L1RhEt followed by b-hydrogen elimination and hy-
dride transfer to the ligand. The observed addition to
the 4-position of the pyridine ring is probably the most
3
Even though the exact formulation of complex 5a
remains unclear, hydrogenation experiments were car-
ried out. When various olefins were added to 5a under
favourable one.
When the reaction of 1a with Et2Zn was performed in
cyclohexane, the result was quite different. Instead of
addition of a hydride to the ligand, a proton was re-
moved from one of the ketimine methyl groups of the
ligand and {L1 ) H}Rh(C2H4) (7a) was formed (Fig. 7).
g-factor
2.017 1.978
1
The complex was positively identified by H NMR.
The doubled set of ligand iPr peaks indicates the
asymmetry of the complex. The vinylic protons of the
deprotonated ketimine methyl group were visible as two
singlets at 3.9 and 4.5 ppm. This time the remaining li-
gand at Rh could be identified as ethene, observed as a
sim
exp
1
Alkyl transfer to the 4-position of the pyridine ring has been
observed in our laboratories in the reaction of L1 with Et3Al. The
results will be published soon: Q. Knijnenburg, P.H.M. Budzelaar,
submitted.
315
325
B [mT]
335
345
2
13C NMR measurements were not possible because of the low
solubility of the complex.
3
It has been calculated that the 2 and 4 positions of the pyridine ring
Fig. 4. EPR spectrum after reaction of L1RhR with H2
(Wx ¼ Wy ¼ Wz ¼ 11:00, Mod. Ampl. ¼ 2 G, Freq. ¼ 9.3045, attn. ¼ 40
dB, T ¼ 40 K). Simulation: gx ¼ 1:9775, gy ¼ gz ¼ 2:0169, ANz ¼ 40
MHz.
in a pyridine diimine vanadium complex are the most positively
charged parts of that complex and therefore the most likely candidates
to accept a hydride [7d].

2950
T.M. Kooistra et al. / Inorganica Chimica Acta 357 (2004) 2945–2952
N
N
N
N
Rh
Me
Ar
Ar
Ar
2a
- CH4
+ X
N
N
H2
N
N
N
N
Rh
Rh
X
Ar
Ar
Ar
Ar
H
Me
5a
N
N
Rh
Ar
H
H
Me
Fig. 5. Possible formation of 5a.
5 LRhEt
3 (L-H)Rh(C2H4) + 2 LRhH2 + 2 C2 H6
7a 5a
N
Fig. 8. Possible formation of 7a from L1RhEt.
N
N
Rh
X'
Ar
Ar
6a
between reactions carried out in benzene/toluene and
cyclohexane could be that, in an arene solvent, ethene is
easily lost from L1RhEt via b-elimination and ligand
displacement, whereas in the absence of such a solvent
ethene can only be lost as ethane, taking a ligand hy-
drogen with it and producing a highly unsaturated Rh
complex. Formation of 7a from 5a and ethene could
similarly be explained via loss of hydrogens as ethane.
However, in the absence of additional evidence these
explanations and the reactions depicted in Figs. 5 and 8
must remain speculative.
Fig. 6. Proposed structure of 6a.
N
N
N
Rh
Ar
Ar
7a
Fig. 7. Structure of 7a.
doublet at 3.5 ppm with a small rhodium coupling of 1.8
Hz.
3.1.4. L2: blocking the pyridine 4-position
€
Complex 7a also decomposed slowly to free ligand
and metallic rhodium and had a low solubility. There-
fore, no 13C NMR spectrum could be obtained. Pre-
sumably, the same ethyl complex is formed initially as in
toluene, but in this case, net dihydrogen elimination
took place instead of hydride transfer to the ligand to
give 6a. The only difference in reaction conditions is the
solvent. Therefore, it seems likely that the coordinating
ability of toluene/benzene plays a role in the formation
of 6a. Surprisingly, treatment of paramagnetic complex
5a with ethene also gave complex 7a.
Rationalising the formation of 7a from the presumed
intermediate L1RhEt is not easy. It is also complicated
by the fact that we do not know the exact stoichiometry
of the reaction, since part of the rhodium could end up
as paramagnetic, NMR-silent 5a. Formally, formation
of 7a from L1RhEt involves loss of H2, but it is unlikely
H2 is lost as such. Assuming that the hydrogen is lost as
ethane, and that 5a is a dihydrogen or dihydride com-
plex (see next section), a possible reaction stoichiometry
compatible with the observations is shown in Fig. 8. If
this is indeed what happens, the reaction must involve
several intermolecular transfers of hydrogen and/or
ethene, similar to that suggested in Fig. 5. The difference
Nuckel reported that substitution of pyridine H4 by a
tert-butyl group significantly increased the solubility of
pyridine diimine rhodium complexes in apolar solvents
[10]. Higher solubility (and stability) of the products
described in the previous sections would be desirable to
obtain more complete NMR characterization of the
products (13C NMR); the additional aliphatic group
could result in better crystallisation behaviour as well.
Moreover, the tert-butyl group causes significant steric
shielding of the pyridine 4-position and might make
hydrogen transfer to the 4-position more difficult.
L2RhMe (2b) was synthesised in two steps from the
ligand and [Rh(C2H4)2Cl]2, similar to 2a. Again, two
equivalents of MeLi were used to achieve complete al-
kylation of the chloride complex and to prevent the
formation of unwanted side products. No crystals suit-
able for X-ray diffraction could be obtained.
When 2b was exposed to H2, a colour change from
1
green to brown was observed. In the H NMR spec-
trum, no signals for either product or starting material
were present. Therefore, it is suspected that a para-
magnetic complex 5b was formed similar to 5a. Unfor-
tunately, also in this case attempted isolation of the
complex was unsuccessful.

T.M. Kooistra et al. / Inorganica Chimica Acta 357 (2004) 2945–2952
2951
-
LDA
C2H4
N
N
N
+ ???
Ar
N
N
N
N
N
N
Rh
Cl
Rh
Rh
X"
Ar
Ar
Ar
Ar
Ar
1b
7b
8b
Fig. 9. Reaction of 1b with LDA and ethene.
If 5b is indeed a complex of Rhð0Þ, it might be possible
to trap it by adding strong p-acceptor ligands (CO, is-
ocyanides). Addition of CO to a sample of 5b produced
a black precipitate; NMR of the remaining solution
showed only signals for free ligand and a signal at 4.46
ppm that we attribute to H2. H2 was not visible in the
NMR spectrum of 5b before reaction. This suggests that
the ligand X in 5a/5b may be dihydrogen or two hy-
drides, which would not be strange for a complex that
was formed in the presence of excess dihydrogen. When
the slightly weaker p-acceptor tBuNC was used, again
loss of ligand and liberation of H2 was observed. In a
separate experiment, addition of CO to 2b also resulted
in loss of the free ligand. Evidently, the strong p-acid
CO does not stabilise LRh complexes.
Reaction of 1b with Et2Zn in toluene initially resulted
in line broadening in the 1H NMR spectrum, but after a
few hours, the reaction was finished and a diamagnetic
complex (7b) had formed. 1H NMR showed that this was
the analogue of 7a formed from 1a in cyclohexane. Two
singlets of vinylic protons at 4.13 and 4.67 ppm indicated
that one of the ketimine methyl groups was deproto-
nated. As in 7a, an ethene molecule is coordinated to the
metal. Apparently, the hydride transfer to the ligand
encountered in toluene and benzene with L1 does not
proceed when the pyridine 4-position is substituted with a
tBu group. Another difference is that 7b is more stable
than 7a. Even after several weeks in solution, only part of
the complex had lost its ligand. This made more extensive
NMR work possible. From a NOESY experiment the
vinylic protons could be assigned. The cis proton showed
a clear contact with the isopropyl group on one of the aryl
rings, while the trans proton had a contact with one of the
In order to confirm the structure of 7b, we attempted
to prepare it via the alternative route of deprotonation
of 1b in the presence of ethene. Addition of one equiv-
alent of an LDA solution to 1b resulted in an immediate
colour change from green to red and in five minutes to
yellow/green. NMR clearly showed that both 7b and the
above-mentioned side product 8b had been formed, but
attempts to separate the complexes were unsuccessful
(see Fig. 9).
4. Conclusions
When the reactivity of LRhCl and LCoCl is com-
pared, the similarity stops after the alkylation to LMR.
In the case of Co, all complexes can be activated with
MAO to become highly active olefin polymerisation
catalysts. It is even likely that they all form the same
active species. All Rh complexes are completely inactive
towards ethene polymerisation, even after reaction with
MAO. The hydrogenation activity of Rh and Co com-
plexes seems similar, but while for Co the active species
appears to be LCoH, for Rh the catalysis is probably
caused by a paramagnetic species (5a). Unexpectedly,
reaction of this paramagnetic species with ethene leads
to a diamagnetic ethene complex containing a deproto-
nated ligand (7a). Hydrogen transfer to the ligand has
also been observed (6a). Thus, the ‘‘chemically non-in-
nocent’’ character of the ligand shows up quite clearly in
its Rh chemistry. For Co, in contrast, we merely see
‘‘electronic non-innocence’’ in the stabilisation of un-
usually low oxidation states [5,6].
One final conclusion from the present work is that the
pyridine diimine ligand, despite being a tridentate li-
gand, is remarkably easily lost from RhI. One possible
reason could be steric: for large metal atoms like Rh, the
ligand ‘‘fit’’ is not very good. But in addition, it seems
likely that the combination of a ligand that easily forms
radical anions with a metal that prefers closed-shell
structures constitutes an unhappy marriage.
protons at the pyridine 3-position. Unfortunately, no 13
NMR spectrum could be obtained.
C
In the reaction of 1b with Et2Zn, formation of a
substantial amount of side product (8b) was observed
(sometimes up to 40%). The typical vinylidene reso-
nances observed for it lead us to suspect that it contains
a doubly deprotonated ligand similar to the one re-
ported by Gambarotta for Mn [8b].
Analogous to the behaviour observed for 5a, treat-
ment of 5b with ethene resulted in very slow formation
of 7b. These observations demonstrate that the com-
plexes of L1 and L2 behave similarly, except that hy-
drogen transfer to the pyridine 4-position is apparently
blocked in L2.
Acknowledgements
This research was sponsored by CW/STW and Basell
Technology Company. We thank Dr. B. de Bruin for
EPR measurements and simulations.

2952
T.M. Kooistra et al. / Inorganica Chimica Acta 357 (2004) 2945–2952
[5] (a) V.C. Gibson, M.J. Humphries, K.T. Tellmann, D.F. Wass,
References
A.J.P. White, D.J. Williams, Chem. Commun. (2001) 2252;
(b) T.M. Kooistra, Q. Knijnenburg, J.M.M. Smits, A.D. Horton,
P.H.M. Budzelaar, A.W. Gal, Angew. Chem., Int. Ed. 40 (2001)
4719.
[1] (a) S.A. Svejda, L.K. Johnson, M. Brookhart, J. Am. Chem. Soc.
121 (1999) 10634;
(b) S.J. McLain, J. Feldman, E.F. McCord, K.H. Gardner, M.F.
Teasley, Macromolecules 31 (1998) 6705;
€
[6] (a) B. de Bruin, E. Bill, E. Bothe, T. Weyermuller, K. Wieghardt,
Inorg. Chem. 39 (2000) 2936;
(c) S. Mecking, L.K. Johnson, L. Wang, M. Brookhart, J. Am.
Chem. Soc. 120 (1998) 888;
(b) P.H.M. Budzelaar, B. de Bruin, A.W. Gal, K. Wieghardt, J.H.
van Lenthe, Inorg. Chem. 40 (2001) 4649;
(d) C.M. Killian, L.K. Johnson, M. Brookhart, Organometallics
16n (1997) 2005;
(c) V.C. Gibson, K.P. Tellmann, M.J. Humphries, D.F. Wass,
Chem. Commun. (2002) 2316.
(e) C.M. Killian, D.J. Tempel, L.K. Johnson, S. Mecking, M.
Brookhart, J. Am. Chem. Soc. 118 (1996) 267.
[2] (a) M. Freemantle, Chem. Eng. News (1998) 11;
(b) G.J.P. Britovsek, V.C. Gibson, B.S. Kimberely, P.J. Madox,
S.J. McTavish, G.A. Solan, A.J.P. White, D.J. Williams, Chem.
Commun. (1998) 849;
[7] (a) G.K.B. Clentsmith, V.C. Gibson, P.B. Hitchcock, B.S.
Kimberley, C.W. Rees, Chem. Commun. (2002) 1498;
(b) I. Khorobkov, S. Gambarotta, G.P.A. Yap, P.H.M. Bud-
zelaar, Organometallics 21 (2002) 3088;
(c) M. Bruce, V.C. Gibson, C. Redshaw, G.A. Solan, A.J.P.
White, D.J. Williams, Chem. Commun. (1998) 2523;
(d) D. Reardon, F. Conan, S. Gambarotta, G. Yap, Q. Wang, J.
Am. Chem. Soc. 121 (1999) 9318.
(c) B.L. Small, M. Brookhart, A.M.A. Bennett, J. Am. Chem. Soc.
120 (1998) 4049;
(d) B.L. Small, M. Brookhart, J. Am. Chem. Soc. 120 (1998)
7143;
[8] (a) D. Reardon, G. Aharonian, S. Gambarotta, G.P.A. Yap,
Organometallics 21 (2002) 786;
(e) G.J.P. Britovsek, M. Bruce, V.C. Gibson, B.G.A. Kimberley,
S. Stroemberg, A.J.P. White, D.J. Williams, J. Am. Chem. Soc.
121 (1999) 8728.
(b) H. Sugiyama, G. Aharonian, S. Gambarotta, G.P.A. Yap,
P.H.M. Budzelaar, J. Am. Chem. Soc. 124 (2002) 12268.
[9] (a)See e.g.: H.F. Haarman, J.M. Ernsting, M. Krananburg, H.
Kooijman, N. Veldman, A.L. Spek, P.W.N.M. van Leeuwen, K.
Vrieze, Organometallics 16 (1997) 887;
[3] G.J.P. Britovsek, G.K.B. Clentsmith, V.C. Gibson, D.M.L.
Goodgame, S.J. McTavish, Q.A. Pankhurst, Catal. Commun. 3
(2002) 207.
(b) E.L. Dias, M. Brookhart, P.S. White, Organometallics 19
(2000) 4995.
[4] (a) L. Deng, P. Margl, T. Ziegler, J. Am. Chem. Soc. 121 (1999)
6479;
€
[10] (a) S. Nuckel, P. Burger, Organometallics 20 (2001) 4345;
(b) P. Burger, personal communication.
[11] R. Cramer, Inorg. Synth. 15 (1974) 14.
(b) E.A.H. Griffiths, G.J.P. Britovsek, V.C. Gibson, I.R. Gould,
Chem. Commun. (1999) 1333;
(c) D.V. Khoroshun;, D.G. Musaev, T. Vreven, K. Morokuma,
Organometallics 20 (2001) 2007;
[12] (a) Q. Knijnenburg et al., to be published;
(b) A.D. Horton, Q. Knijnenburg, H. van der Heijden, P.H.M.
Budzelaar, A.W. Gal, WO 030442131 (2003).
~
(d) J. Ramos, V. Cruz, A. Munoz-Escalona, J. Martınez-Salazar,
ꢀ
Polymer 43 (2002) 3635.