Late Transition Metal Complexes of Mixed NHC / Phenolate Tripodal Ligands for Small Molecule Activation Komplexe später Übergangsmetalle tripodaler, gemischter NHC/ Phenolat Liganden zur Aktivierung kleiner Moleküle Der Naturwissenschaftlichen Fakultät der Friedrich-Alexander-Universität Erlangen-Nürnberg zur Erlangung des Doktorgrades Dr. rer. nat. vorgelegt von Martina Käß aus Schwabmünchen
267
Embed
Late Transition Metal Complexes of Mixed NHC / Phenolate ... · Late Transition Metal Complexes of Mixed NHC / Phenolate Tripodal Ligands for Small Molecule Activation Komplexe später
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Late Transition Metal Complexes of Mixed
NHC / Phenolate Tripodal Ligands for Small
Molecule Activation
Komplexe später Übergangsmetalle tripodaler, gemischter
NHC/ Phenolat Liganden zur Aktivierung kleiner Moleküle
Der Naturwissenschaftlichen Fakultät
der Friedrich-Alexander-Universität
Erlangen-Nürnberg
zur
Erlangung des Doktorgrades Dr. rer. nat.
vorgelegt von
Martina Käß
aus Schwabmünchen
Als Dissertation genehmigt
von der Naturwissenschaftlichen Fakultät
der Friedrich-Alexander-Universität Erlangen-Nürnberg
Tag der mündlichen Prüfung: 13.06.2014
Vorsitzender des Promotionsorgans: Prof. Dr. Johannes Barth
Gutachter: Prof. Dr. Karsten Meyer
Prof. Dr. Nicolai Burzlaff
Experience is what you get when you didn’t get what you wanted.
This thesis concerns itself with tripodal, mixed NHC/phenolate ligands, their coordination
chemistry, and the possible use of their complexes in small molecule activation. This
introductory chapter will first map out some of the biological background for the work,
then go on to discuss the specific properties of tripodal ligands. Subsequently, the two
different classes of binding groups – N-heterocyclic carbenes and phenolates – will be
elaborated on in more detail, to give a summary of the history and use of the respective
ligand classes, and the inherent properties of these binding sites. Finally, this background
on the state of the art is linked to the questions I tried to answer with this thesis.
1.1 Motivation and Bio-Inorganic Background
Certain small molecules such as N2 and CO2 possess extremely high thermodynamic
stability and hence low reactivity. The bond dissociation energy for the N≡N-triple bond,
for instance, is 945 kJ/mol high, and the highest dissociation energy is found in carbon
monoxide C≡O (1077 kJ/mol).[1] It is therefore a challenge to coordinate these molecules to
a metal center and activate them for further reactions. Biological fixation of atmospheric
dinitrogen is the basis for all nitrogen containing molecules in life forms, from DNA to
amino acids. And the exhaust gas CO2, of which an estimated 33 billion tons are emitted
into the atmosphere every year,[2] may become a source for sustainable fuel and chemical
production – if ways are found to bring this molecule, which is at the deep end of the
thermodynamic well, back into the industrial process through reduction or
functionalization.[3-5]
Nature has developed some very effective enzymes for these tasks. Nitrogen fixation is
catalyzed by nitrogenase, a protein with a reactive FeMo cluster at its center[6-7] that is
found in certain bacteria and archaea.[8] Some of the proposed intermediates during the
transformation process are high-valent metal species with metal-nitrogen multiple bonds.[9]
In the corresponding industrial process (Haber Bosch) atmospheric N2 is converted to
136 million tons of ammonia per year.[10] The cleavage of the N2-triple bond also proceeds
via imido and nitrido species that have been observed spectroscopically on the iron
catalyst’s surface.[11-12]
2
Nature’s way of CO2 fixation is photosynthesis, which turns CO2 and water into
carbohydrates. At the other end of that process, dioxygen is formed by a light-driven four-
electron oxidation of water. The dioxygen is released in photosystem II from the oxygen
evolving complex (OEC), an enzyme with a Mn4CaO5 cluster at its core.[13] The
mechanism of O–O bond formation is still under debate. There are mainly two reaction
pathways discussed in literature today: Firstly, an acid/base pathway (AB, or nucleophilic
attack mechanism), in which a water molecule (the nucleophilic oxygen or base) attacks a
terminal oxo-moiety (the electrophilic oxygen or acid); secondly, a radical coupling
mechanism (RC) between two terminal oxo species.[14] While there is experimental
evidence for the AB mechanism in Mn model systems that perform O–O bond
formation,[15-16] more recent computational studies using the latest OEC crystal structure[13]
favor an oxo/oxyl radical coupling mechanism.[17-18]
The development and characterization of well-defined model complexes of species with
metal-to-ligand multiple bonds are vital to aid in the elucidation of these processes and
their mechanism.[19] Functional models or reactivity studies may also promote a deeper
understanding of the underlying principles. With deeper insights into reaction pathways at
reactive metals centers, not only may Nature be better understood, but more efficient
catalysts may be developed for industrial processes. The aforementioned Haber-Bosch
process consumes vast amounts of energy and natural gas, as it requires high pressure and
high temperatures as well as dihydrogen, which is in turn produced in the energy-intensive
water gas shift reaction that also sets free carbon dioxide. Catalysts that generate ammonia
from N2 at milder conditions may help preserve large amounts of energy – Nature’s
nitrogenase is still the best role model for this.
Furthermore, complexes with oxo-, imido-, and nitrido species may be used in catalytic
atom- and group transfer reactions, i.e. for epoxidation and aziridination reactions, or even
direct C–H-bond amination.[20] For all these reasons, the synthesis of species with
metal-ligand multiple bonds is an attractive research goal. For late transition metal
complexes, tripodal ligands have emerged as a most suitable class of ligands to stabilize
such M=E and M≡E multiple bonds.
3
1.2 Tripodal Ligands
Polydentate ligands give more robust complexes than their monomeric counterparts, owing
to the (kinetic and thermodynamic) chelating effect. Thus, tripodal ligands with their three
(or more) binding sites often form stable complexes with a large variety of metal ions. The
steric bulk around the metal center depends on the nature of the coordinating groups, but
can often be tuned conveniently by exchanging their substituents, and in that way bulky
ligands may be easily prepared. The anchoring units of tripodal ligands – usually B, C, or N
– can provide further structural and electronic flexibility, and they can effectively block
undesired side- and decomposition reactions by shielding one side of the reactive metal
center entirely.
As a result of its trigonal nature, a tripodal ligand enforces ‘facial’ (as opposed to
‘T-shaped’ or ‘meridional’) binding. This can be used to mimic a variety of non-heme
metallo-enzymes, many of which contain a triad of amino acid side chains that binds
facially to the metal center.[21-23] In contrast, other 3-coordinate geometries, like the
‘T-shaped’ binding of pincer-type ligands for instance, have little to no significant
biological relevance.
All these beneficial characteristics of tripodal ligands as well as their complexes’ electronic
structure (see below) have promoted the design and development of numerous tripodal
ligand systems.
1.2.1 Tripodal Ligands in Literature
One old and widespread class of tripodal ligands are tris(pyrazolyl)borates (Tp)
(compound A in Scheme 1)[24-25] and their derivatives. They were first published by
Trofimenko in 1966,[26] who named them “scorpionate ligands”,[27] evoking the picture of a
scorpion that grabs its prey (the metal ion) with its two pincers before reaching over the
plane to attack with its stinger (the third ligand arm).1 The steric and electronic properties
1 The Tp ligands are usually referred to as “scorpionate ligands” in literature, as are some other ligand
systems mentioned here (e.g. Tm). The differentiation between scorpionates and tripodal ligands is nebulous,
however, and the question of whether a ligand system is called a “scorpionate” or simply a “tripodal ligand”
in literature seems to be a question of the school or chemical heritage of a publications’ author or ligands’
creator, rather than a logical classification or distinction between ligands.
4
of the pyrazolyl donor can be modified by changing its 3- and 5-substituents (R and R’).[24-
25] Numerous derivatives of the Tp ligand and of the related tris(pyrazolyl)methane (tpm,
B) have been developed, although the synthesis of the latter is more challenging.[28-31]
Also very popular and effective, the tris(amido)amine ligands, based on inexpensive tris(2-
aminoethyl)amine, have been intensively studied.[32] Schrock and coworkers employed a
bulky tris(amido)amine to catalytically reduce dinitrogen to ammonia.[33] They isolated and
BNN
N N
R
H
R
N
N
R
R'R'R' C
NN
N N
R
H
R
N
N
R
R'R'R'
tris(pyrazolyl)borate, Tp
B
P PP
FeIV
N
N
N NMoIIIN
iPr iPr
iPr
iPr
iPriPr
HIPT
HIPT
N
N
N
N NFeIIIN
O
N
O
N
ON H
HH O
2-
-
A
tris(pyrazolyl)methane
B
a tris(amido)amine complex
C
tris(ureayl-amido)amine,
D
B
S SS
NiII
Fe-oxo complex
AdAd
O O
tris(thiomethyl)borate,
E
Ni-superoxo complex
tris(phosphino)borate,
F
Fe-nitrido complex
BNN
N N
R
H
RN
N
R
tris(imidazol-2-thione)borate, Tm
-
H
SS
S
N
O O
N
N
O
tris(oxazoline)amine
G
N
R R'
R R'
R R'
Scheme 1. Several representative tripodal ligands and complexes (R, R’ = various alkyl and aryl
substituents).
5
characterized a number of potential intermediates,[34-37] one of which is shown in Scheme 1
(compound C, HITP = hexa-iso-propyl-terphenyl). Borovik and coworkers prepared
terminal iron(III) oxo complex D using a ureayl-functionalized tris(amido)amine ligand,
which stabilizes the oxo-ligand through hydrogen bonding.[38-39]
In contrast to the “hard” amido donor ligands, tripodal ligands bearing ‘‘soft’’ donor atoms,
such as sulfur[40-41] and phosphorus, are more suitable for stabilizing electron-rich low-
valent and late transition metal centers. Thus, nickel complexes of a sulphur-borane tripod
were synthesized, as was its side-on dioxygen adduct identified as a Ni(II)-superoxo
complex (E).[40] Peters and coworkers demonstrated the high potential of tripodal ligands in
small molecule activation by utilizing a tris(phosphino)borane[42] to synthesize iron nitrido
species (F),[43] as well as terminal Co(III) imido[44] and several Fe(III) imido[45-46]
complexes.
Further tripodal ligand systems included in Scheme 1 are the tris(oxazolines), which have
been developed with either methane-, nitrogen- (G), cyclopropane- or mesityl-anchoring
units,[47] and tris(imidazol-2-thione) ligands (H, TmR: hydro-tris(R-thimazolyl)borate).[48]
Tripodal NHC Ligands
Tripodal NHC ligand systems have so far been synthesized with mesityl-, borane-,
methane- and nitrogen-anchors. The mesityl-anchored ligand I (Scheme 2) was isolated by
Dias and Jin, but no metal complexation was achieved.[49] Hu and Meyer were able to
introduce the exceptionally large Tl(I)-ion into the cavity, but it was concluded that the
ligands’ cavity was too large (and rigid) to allow the complexation of any transition metal
ion.[50]
As analogue to the common Tp ligands, Fehlhammer synthesized a tris(carbene)borate.[51]
However, this ligand formed complexes with two ligand molecules chelating one metal
center, resulting in coordinatively saturated hexakis(carbene) complexes that did not bind
any further ligands (see complex J in Scheme 2), which decisively limits their applicability
in small molecule activation.[51-53] By derivatizing this ligand with larger substituents on the
boron anchor, and more importantly on the NHCs, Smith was not only able to isolate 1:1
complexes,[54-56] but also synthesize iron- and cobalt imido-complexes,[57-58] as well as iron-
nitridos[59-60] that form ammonia in the presence of an H-atom donor[61] (Scheme 3).
6
N
N
NN
N
N
BNN
N N
R
H
RN
N
R
BN N
NN
R
H
RN
N
RMII
M = Co, Rh, Fe
2+
I J R = Me, Et
N
N
NC
N
RC
N
MN
C
N
R
R
C
N
NC
N
RC
N
MN
C
N
R
R
hypothetical
12
3
456
7
8 12
3
45
6
1:1 TIMER metal complexTIMENR metal complex
Scheme 2. Tripodal tris(carbene) ligands and complexes.
BN
N
N
N N
N
Fe
N
NN
NC
NC
N
FeIV NC
N
R
RR
N
+
NN
NC
NC
N
Mn NC
N
N
n+
MnIII (n = 0),
MnIV (n = 1),
MnV (n = 2)
R = H, Me
n+
FeIV (n = 0),
FeV (n = 1)
Scheme 3. Nitrido complexes from the groups of Smith (left) and Meyer (middle and right).
7
The methane-anchored NHC tripodal ligand TIMER (1,1,1-tris(3-R-imidazol-2-ylidene)-
(methyl)ethane, R=me, tBu) by Hu and Meyer formed complexes with group 11 metals, but
never in a 1:1 ligand : metal ratio.[62-64] This may be explained insofar as a hypothetical 1:1
TIMER : metal complex would contain three eight-membered metalla-rings, with much
lower stability than the more commonly observed five- or six-membered rings (see
Scheme 2).[64] Therefore, Hu and Meyer sought to incorporate a coordinating atom at the
anchoring position of the carbene tripod, which would result in more favorable six-
membered rings. The N-anchored tris(carbene) ligand TIMENR (tris[2-(3-R-imidazol-2-
ylidene)ethyl]amine)[64] finally gave the desired 1:1 complexes with, among others, Mn, Fe,
Co, Cu,[65] and Ni.[66]
TIMENR (R = mesityl, 2,6-xylyl) has enabled the stabilization and full characterization of
terminal Fe(IV) nitrido complexes, and these were the first crystallographically studied
mononuclear iron-nitrides.[67] Recently, a series of TIMENR (R = 2,6-xylyl) Mn nitrides in
different oxidation states (III, IV, and V) has also been synthesized.[68] Their reactivity is
currently investigated. In the cobalt peroxo complex of the TIMENXyl ligand, two of the
carbene arms are pushed apart, giving the complex an overall distorted octahedral
symmetry, thus allowing for side access for organic substrates.[69] Consequently, the
complex was employed in O-atom transfer chemistry to organic substrates. In contrast, the
three-fold symmetrical iron nitrido and cobalt imido complexes of TIMENR (R = mesityl,
2,6-xylyl) did not undergo any atom or group transfer chemistry. Instead it could be shown
that the Co(III) imido and the one-electron oxidized Fe(V) nitrido intermediate insert the
RN2– and N3– ligands into the metal carbene bond, forming bis(carbene) imine species
(Scheme 4).[70-71]
Therefore, reduction of the steric pressure of the chelating ligand was desired, to open up
the reactive cavity and allow side access of substrates to the functional entity. The first
strategy to reduce steric pressure was to remove the ortho-CH3-groups of the mesityl-/
xylyl-substituents on the imidazolylidenes. However, the introduction of reactive
hydrogens in the ortho positions leads to unexpected new reactivity, resulting in C–H bond
activation and three-fold metalation of the ligand.[72] Subsequently, side access by
exchanging one or two carbene arms with phenolates was sought (section 1.5 Objectives).
8
NN
NC
NC
N
M NC
N
R'
R'R'
N
R
N
N
NC
NC
N
MNC
N
R'
R'
N
R'
R
R' = H, Me
a) M = Fe, Co,
b) M = Fe,
R = Ar
N = nitride
R' = H, Me
a) M = Fe, Co,
b) M = Fe,
R = Ar
R = H Scheme 4. Insertion reactions of TIMENR imido and nitrido complexes.
Tris(phenolates)
Another N-anchored tripodal ligand system routinely employed in the Meyer group
laboratories[73-75] is the tris(phenolate) ligand ((R,R’ArO)3N)3– (trianion of tris[(3,5-R,R’-2-
hydroxyphenyl)methyl]amine)[76] (Scheme 5). A tris(phenolate) derivative of this ligand
system with methyl substituents in the ortho and para position of the phenol was employed
by Kleij and coworkers in the catalytic cycloaddition of CO2 to epoxides and oxiranes with
iron.[77] The tris(phenolate) ligand derivative ((Ad,MeArO)3N)3–, developed in our lab, has
recently been used to activate CO2 and other heteroallenes at reactive uranium coordination
complexes.[78]
It is worth mentioning that the equivalent tris(thiolate) system has been developed about 2
decades ago,[79] and particularly its chemistry with iron and molybdenum has been explored
in the hope of mimicking as far as possible the FeMo-cofactor of nitrogenase.[80] Very
recently, even a corresponding tris(stibine) (i.e. with three antimony binding groups) has
been prepared and its coordination to Fe and Mn probed.[81]
N
OR
R'3
N
SR
R'3
N
SbR
R'3
Scheme 5. Tris(phenolate) (left), tris(thiophenolate) (middle) and tris(stibine) N-anchored ligands.
9
Enantiomery in Tripods
The TIMENR ligand (and probably other tripods) forms helical enantiomers upon
coordination, which are observed however only in the solid state (i.e. by X-ray single
crystal analysis). In solution, the system can be considered dynamic, with a rapid exchange
between the two enantiomers resulting time-averaged in C3v symmetry.
In search of catalysts for enantioselective catalysis, a number of chiral, enantiomerically
pure tripodal ligands have been synthesized, including derivatives of tris(oxazolines),[47]
tris(pyrazolyls) and tris(phosphanes).[82] While bidentate C2-symmetric ligands reduce the
number of possible diastereomers in catalytic intermediates for square planar complex
geometries, C3-symmetry can do the same for octahedral intermediates.[47, 82]
Mixed Tripodal Ligands
All tripodal ligands mentioned so far are also C3 symmetric, with three equivalent binding
groups, but a number of tripodal ligands with mixed ligand arms are known.
Of the borane-anchored „scorpionate“ ligands, mixtures between pyrazolyl and imidazole-
2-thione ligand arms[48], and for the carbon-anchor, between pyrazolyl and carboxylate
binding groups[22] have been published.
Some ligand arms that are easily mixed on a nitrogen anchoring unit are N-donors like
pyridines, imidazoles, benzimidazoles, or amines, and O-donors like phenolates or
alkoxides, as has been nicely demonstrated by Palaniandavar and coworkers.[83] Most
impressively, a monomeric Mn complex of an N-anchored, mixed ureayl and pyridyl
ligand has been shown by Borovik and coworkers to catalytically reduce dioxygen to
water.[84]
So far, and to the best of my knowledge, no tripodal mixed NHC / phenolate systems have
been reported.
1.2.2 Geometry and Steric Bulk
For a tripodal ligand system, the sterics around the complex’s metal center depend to a
large extent on the nature of the binding groups. Scheme 6 demonstrates that for
tris(amides) and –(phosphanes), the organic substituents R are directly bound to the
coordinating atom, and therefore point away from the metal center. This leaves the reactive
site wide open, encouraging dinuclear decomposition pathways. Accordingly, the iron(IV)
10
nitride of Betley and Peters dimerizes and could not be characterized by X-ray
crystallography: The single crystals contain merely the dinuclear, dinitrogen-bridged
coupling product.[43]
Similarly, Yandulov and Schrock had to introduce three exceedingly bulky hexa-isopropyl
terphenyl substituents at the tris(amido)amine ligand in order to prevent dimerization of
their Mo-nitride intermediate.[33] The synthesis of these extremely bulky ligand derivatives
is often challenging and time-consuming.
In contrast, the sterics of tripodal NHC ligands are controlled by the substituents at the
imidazole N3 position, which can align nearly perpendicular to the metal-carbene
coordination plane. Even for less bulky alkyl or aryl substituents, this results in a narrow,
well-protected cylindrical cavity around the reactive center, which allows axial access for
small molecules and is able to stabilize reactive species like the terminal nitride ligands
shown above in Scheme 3.
M
P PP
B
Ph
R
R
R
R
RR
M
C C
N N N
B
Ph
NMN
N
N
R
R
R
NM
N
N
N
N
C
C
N
C
N
R
R
R
NN C
N
RR
R
Scheme 6. Sterics around the reactive metal center for tripodal ligands with different binding
groups; arrows pointing in the direction of the steric bulk created by the substituents R (adapted and
expanded from Hu et al.[64]).
11
The comparison between Smith’s borane-ligand and Meyer’s TIMENR demonstrates that
the steric bulk can also be strongly influenced by the anchor or the “bridge length” between
anchor and binding groups: Both ligands are tris(carbenes), but in Smith’s case, the NHCs
are directly bound to the boron anchor. Thus, the carbenes are forced to coordinate at a
flatter angle to each other, which causes the substituents R to point somewhat away from
the reactive center.
1.2.3 Electronic Structure
Tripodal ligands provide a powerful platform for small molecule activation chemistry,[32-33,
35, 84-87] as has been highlighted most impressively by Peter’s synthesis of Fe(III) imido,[45-
46] terminal Co(III) imido,[44] and iron nitrido species,[43] plus further examples given in
section 1.2.1 (see Schemes Scheme 1 and 3). The ligand field splitting resulting from their
trigonal coordination environment is suitable for the stabilization of highly unusual metal-
ligand multiple bonds, even for relatively electron rich late transition metals,[87-88] which
will now be illustrated and discussed by means of the d-orbital splitting diagram in
Scheme 7 (see also reviews by Nocera[88] and Peters[87]):
Scheme 7. d-Orbital splitting patterns for complexes with metal-ligand multiple bonds M=E and
M≡E in C4 and C3 symmetric environments.
12
In an octahedral or square pyramidal complex with only σ donor ligands, one finds three
degenerate, non-bonding orbitals (dxy, dxz, and dyz) and two anti-bonding ones (dz2, dx2-y2).
Through interactions with π-donors, some of the non-bonding orbitals may become
antibonding. For strongly π-donating ligands, such as oxos or nitridos, this results in the
1+2+1+1 splitting pattern ( 7a) ) first introduced by Ballhausen and Gray for the vanadyl
ion in VO(H2O)5.[89] This pattern has been used by Gray for trans-dioxo and nitrido
complexes of Tc(V), Re(V),[90-91] and Os(VI)[92], and adapted for many similar complexes.
The more covalent the ligand-to-metal multiple bond becomes, the more the dxz and dyz
involved in the π-bonding are destabilized. For trans-[(cyclam)M(N)(X)] complexes
(M = Cr, Mn; X = µ-N3, CH3CN), a 1+3+1 orbital splitting with almost degenerate dx2-y2,
dyz, and dxz orbitals has been deduced,[93-94] and finally in [Cr(N)Cl4]2–, the π* orbitals are
even higher in energy than the dx2-y2, leading to a 1+1+2+1 splitting ( 7b) ).[95] In the latter
complex, the π-perturbation from the nitride far outweighs the σ contribution of the
chlorides, which puts the dx2-y2 and dxy orbitals far below the anti-bonding π* involved in
the triple bond to N, giving the complex a d-orbital splitting more comparable to a linear
geometry as reference system than to a distorted octahedron.
In either case, 7a) and 7b), only one non-bonding orbital remains which can house only two
d-electrons, rendering it obvious that this geometry is unfavorable for the higher electron
count of low oxidation states or late transition metals as the population of anti-bonding
orbitals would destabilize the metal-to-ligand multiple bond. This is made evident in the
difference in stability of the three nitrides sketched in Scheme 8: While the Cr(V) and
Mn(V) nitrides (d1 and d2) are stable at RT, and therefore characterized
crystallographically,[93, 96] the iron(V) nitride (d3) can only be generated at very low
temperatures and was studied in frozen matrix by low-temperature Mößbauer and EPR
spectroscopy.[97]
N N
N N
MV
N
L
HH
H H
M = Cr, Mn L = NCMe
M = Fe L = N3
Scheme 8. Cyclam metal-nitrido complexes by Meyer et.al.
13
In a trigonal symmetry, in contrast, the orbitals lying in the xy-plane remain non-bonding
(and degenerate). For an M=ER double π-bond as found in imido-complexes, three non-
bonding orbitals are available (which may even be strictly degenerate) for up to six
d-electrons ( 7c) ).[98] With an oxo or nitrido triple bond, wherein the dz2 is strongly
destabilized through σ-bonding, two non-bonding orbitals still remain to accommodate four
d-electrons. DFT calculations verified this 2+1+2 pattern ( 7d) ) nicely for Vogel’s
[(TIMENR)FeIV(N)] (R = Xyl, Mes).[67] It has also been stated that two frontier π–orbitals
that are strictly degenerate are favorable for metal–ligand multi-bond formation.[32]
In conclusion, the electronic environment created by C3 symmetric ligands can greatly aid
in the synthesis of species with metal-ligand multiple bonds, like terminal imido, oxo and
nitrido species, especially in mid- to late first-row transition metal complexes.
1.3 Carbenes
A carbene is defined as a molecule containing a neutral, divalent carbon atom with only six
valence-shell electrons, in which the carbon atom possesses two covalent bonds to adjacent
atoms. The history of their research can be traced back almost 200 years,[99] and carbene
chemistry today still represents an exciting and rapidly developing area of research.
History and Use in Catalysis
Isolable carbenes have been sought after since the first half of the 19th century.[99-102] The
first notable experiments were conducted by Dumas,[103] who reacted methanol with
dehydrating agents, like phosphorous pentoxide or conc. sulphuric acid, in the hope of
freeing the CH2 unit. His efforts and many others[104-107] failed, however, until at last
carbenes were thought of as unstable, fleeting intermediates by most chemists and treated
as such in standard textbooks. At best, it was thought, they could be stabilized in the form
of metal complexes.
Meanwhile, carbenes were introduced by Doering as synthetic intermediates into organic
chemistry in 1954[108] and by Fischer into organometallic chemistry in 1964,[109] and these
intriguing species became involved in many reactions of high synthetic interest.
14
NC
N
Ph
Ph
H
CCl3
∆∆∆∆
- HCCl3
2N
CN
Ph
Ph
N
N
Ph
Ph
CN
CN
Ph
Ph
2
Scheme 9. Synthesis of an ene-tetraamine (“carbene-dimer) by α-elimination from an imidazolidin.
Wanzlick tried to generate a free carbene through α-elimination of chloroform from the
corresponding imidazolidine compound (Scheme 9),[110] but only the dimer could be
isolated. In 1968, Öfele[111] and Wanzlick[112] independently synthesized chromium and
mercury complexes of imidazolydenes (N-heterocyclic carbenes, or NHCs). Finally, after
renewed efforts,[102] Arduengo was able to isolate, characterize, and even
crystallographically analyze a free carbene (Scheme 10).[113]
Arduengo’s discovery was followed by a sheer explosion of new isolable carbenes. Today,
even a number of non-cyclic stable carbenes have been isolated.[114-116] But NHCs in
particular began to play an important role in transition metal catalysis, as several extremely
active catalysts were developed, perhaps most prominently illustrated by the second-
generation olefin metathesis catalysts developed by Grubbs[117] and Nolan,[118-120] or by the
cross-coupling catalysts introduced by Organ and commercialized by Aldrich.[121]
Catalytic applications of NHC complexes have been extensively reviewed in the last
decade.[122-133] The ongoing popularity and explosive development around this fascinating
class of molecules is demonstrated by dozens of review articles published each year on
carbenes in all fields of chemistry.
N
NH
- NaCl, -H2
(THF)
NaH
[DMSO]
NC
N
Cl
Scheme 10. Synthesis of the first isolated, stable carbene by deprotonation of an imidazolium
chloride, sterically encumbered by bulky adamantyl substituents.[113]
15
Properties and Electronic Structure
Depending on their hybridization, carbenes can be either linear or bent. The simplest of all
carbenes, methylene CH2, is a linear carbene with triplet ground state. The linear geometry
implies sp hybridization, which leaves two degenerate p orbitals for the two non-bonding
electrons, resulting in a triplet ground state. This geometry is adopted only when the
substituents are lower in electronegativity than the carbon atom. Most carbenes possess a
bent coordination mode (sp2 hybridization), and the exact angle between substituents
depends on their electronegativity (which influences the degree of hybridization) and, in
part, on their steric demands (for cyclic carbenes, this includes the ring size).
Scheme 11. Possible electronic configurations of a carbene.[101]
Scheme 11 gives an overview of the possible electron configurations for an sp2 hybridized
carbene. The non-bonding orbitals are one of the hybrid orbitals and one p orbital,
conventionally denoted as the σ and pπ orbital, respectively. Four electronic configurations
can be envisioned:[101] A singlet with both electrons located in the hybrid orbital (σ2), a
triplet ground state (σ1pπ1), an energetically higher singlet state with both electrons in the
p orbital (σ0pπ2), and an excited singlet state with one electron in each orbital (σ1pπ
1) but
with antiparallel spin.
The energy gap between σ and pπ necessary to induce a singlet ground state (σ2) vs. a triplet
(σ1pπ1) was determined by Hoffmann to be 2 eV.[134] The energies of the carbenes’ frontier
orbitals depend on the substituents, which can be explained through inductive (Scheme 12)
and mesomeric (Scheme 13) effects:[101] The inductive effect involves the σ orbital, which
is stabilized by electronegative substituents, leading to larger σ–pπ energy gap and singlet
ground state, whereas substituents with +I–effect decrease the energy gap and,
consequently, favor the triplet ground state. The mesomeric effect engages the pπ orbital,
which is destabilized by π-electron-donating substituents and stabilized by π-electron-
withdrawing substituents.
16
carbene substituent
2px,y,z pπ (b1)
2s
b2
σ (a1)
a1
carbene substituent
2px,y,z pπ (b1)
2s b2
σ (a1)
a1
C C
Scheme 12. Perturbation orbital diagram illustrating the inductive effect of σ-donating (left) and
σ-withdrawing substituents (right) on the carbene orbitals.[101]
pπ
pπ (b1)
σ
a2
σ
b1
CX X
CX X
δ−
1/2 δ+ 1/2 δ+
px pπpy
σ
CZ Z CZ Z
δ+
1/2 δ− 1/2 δ−
σ
pxpy
CX Z CX Z
δ+ δ−
pπ
Scheme 13. Perturbation orbital diagram showing the mesomeric effect on the carbene frontier
orbitals; X = π-donating substituent, Z = π-withdrawing substituent.[101]
While the substituents determine the carbene’s electronic structure, the carbene’s ground
state multiplicity strongly influences its reactivity: Since singlet carbenes possess one filled
HOMO and one empty LUMO, they can be considered ambiphilic, whereas triplet carbenes
possess diradical character.[135]
17
NHCs have replaced phosphines in many areas, for they are temperature stable and do not
oxidize like their phosphine counterparts, which greatly facilitates preparation and storage.
The excellent performance of NHC catalysts derives not exclusively from the pronounced
σ-donor properties of NHC ligands, but also from the possibility of π-back-bonding within
the metal-carbene unit,[136-139] which enables them to stabilize both high and low oxidation
states formed in catalysis resulting in higher turnover numbers and longer catalyst
lifetimes.[140]
Furthermore, they provide one of the few series of ligands that support homogeneous
catalysis and for which, at the same time, steric and electronic parameters can be
extensively tuned.[140-142] Serviceably, while phosphines PR3 have only one variable (their
substituent R) which influences both their steric and electronic properties, an NHCs’ steric
and electronic tuning is separable to a greater extent: Their sterics are predominantly
controlled by substituents on the atoms adjacent to the carbene center (usually the N-
atoms);[143] electronic effects are also influenced by substituents on the other ring-atoms,
and by the nature of the azole ring and the ring position to which the metal is attached.[144]
Classification
Classically, metal carbene complexes have been divided into Fischer and Schrock
carbenes.[145-147] Fischer carbenes are electrophilic and are usually found with mid- to late
transition metals in low oxidation states, and also π-acceptor ligands on the metal and
π-donor substituents on the carbene ligand (typically OR or NR2) that give them further
stability. The nucleophilic Schrock carbenes, on the other hand, are found with early
transistion metals in high oxidations states, with π-donor metals, and normally without
heteroatom-substituents on the carbene carbon atom. Consequently, metal-carbene bonding
in Fischer carbenes is usually described by help of the Dewar-Chatt-Duncanson model,[148-
149] with carbene–metal σ–donation and metal–carbene π–backbonding, whereas the
situation in Schrock carbenes is more aptly described as covalent double-bonding between
a triplet carbene and a triplet metal fragment.[145]
Since the introduction of N-heterocyclic carbene (NHC) complexes, however, the
Fischer/Schrock classification needs to be revisited and/or extended because of the very
different electronic character of these ligands. On first glance, their heteroatom-substituents
18
would put them in the group of the Fischer carbenes. Accordingly, the –I and +M-effect of
the heteroatoms deliver ideal stabilization of the singlet ground state. Nevertheless, “One of
the most characteristic features of NHCs is their extraordinary electron richness.”[150]
Finally, in addition to classic N-heterocyclic carbenes, a growing number of carbenes with
unusual coordination mode have been reported in the last decade,[141, 151] as has recently
been excellently reviewed by Crabtree:[144] abnormal NHCs (aNHC), alternatively called
mesoionic carbenes (MIC), and remote NHCs (rNHC) (see Scheme 14). For free aNHCs,
no six electron structure can be drawn, but one is obliged to assign formal charges to the
non-binding NHCs (thus the name MIC). Therefore, aNHCs are not strictly carbenes
according to the terms’ definition. However, upon binding, both types of NHCs, normal
and abnormal, form compounds with similar behavior and properties, and so the NHCs
nomenclature “holds for aNHCs because it emphasizes the many factors that are common
to both classes of ligand. The distinction between mesoionic and non-mesoionic carbenes
in the free state is largely lost on binding.”[144] Lastly, in rNHCs, the carbene carbon is not
situated α to any heteroatom. An rNHC can be a normal NHC or an aNHC.
NC
N R'R
C
N N R'R C N
Scheme 14. Normal NHC (nNHC, left), abnormal NHC (aNHC, middle) and remote NHC (rNHC,
right). The rNHC in this example is also a normal carbene.
1.4 Phenolate Ligands
Aryloxides look back on a long history of coordination to transition metals.[152] A large
variety of lanthanide[153] and actinide[154] phenolate complexes have been characterized as
well, since metals as diverse as vanadium, iron or uranium[155-156] form strong metal-
oxygen bonds.
Electronic Versatility
Owing to the two lone pairs on the oxygen atom, phenols possess the potential to donate
either one, three, or five electrons. They can do so through π-donation to a single metal
19
center or, in a bridging coordination mode, to different metal atoms. Even as a pure σ-
donor, their strong inductive effect provides stabilization for higher metal oxidation states.
Redox Non-Innocence
In the last two decades, phenolates have become well-established as redox non-innocent
ligands.[157-158], 2 Studies on the phenolate ligands’ redox-activity have been spurred by the
discovery that the reactive center of several metalloenzymes contains phenoxyl radicals. A
tyrosyl radical can be found, for instance, in class I ribonucleotide reductase and
prostaglandin endoperoxide synthase.[159-160] A well-studied case is the reactive center of
galactose oxidase (GO), where a phenoxyl radical is coordinated to a single copper
atom.[161-162] The active form of GO had been regarded as a copper(III) species,[163-164] until
Raman spectroscopy revealed the formation of the phenoxyl radical species and the Cu(II)-
phenoxyl radical bond.[161-162, 165-166]
Since the radical coordination in the active center of enzymes was discovered, many metal-
phenoxyl radical complexes have been synthesized and characterized to reach a better
understanding of the enzymes’ mechanisms and the properties of metal complexes with the
coordinated phenoxyl radicals.[167-168] Copper-phenoxyl complexes have been studied in
particular,[169-170] due to their kinship with the GO center, but the first synthetic phenoxyl
radical complex formation was discovered by Wieghardt et al. after chemical and
photochemical oxidations of Fe(III)-tris(phenolate)-tacn complexes (see Scheme 15).[171]
The historic confusion about the GO reactive center demonstrates nicely that a formally
assignable oxidation number does not have to be the correct one. In the presence of
potentially redox non-innocent ligands, one must always keep in mind that a redox-event
may take place at either the ligand – leading to an open shell radical ligand – or the metal
center (Scheme 16). Whether the oxidation of a metal-phenolate complex really leads to a
metal-phenoxyl or a high-valent metal-phenolate, which are isoelectronic to each other, can
only be determined by suitable spectroscopic methods.
2 Catechols, which are oxidized to semiquinones, have been considered redox non-innocent for a longer time,
while simple phenols have been recognized as redox-active later on.
20
chemical oxidation
NFe
N
N
O
tBu
O
tBu
OtBu
NFe
N
N
O
tBu
O
tBu
OtBu
+
photochem. oxidation
irradiation at 254 nm
Scheme 15. Iron(III)-phenoxyl radical formation through chemical or photochemical oxidations of
an iron(III) 1,4,7-tris(3-tert-butyl-2-oxybenzyl)-1,4,7-triazacyclononane complex (from [170],
Scheme adapted from [171]).
To facilitate the study of redox-behavior of the phenolate ligand, complexes with redox-
inactive, usually diamagnetic metal centers can be prepared for comparison of experimental
data, Zn(II) [172] or Ga(III)[74] complexes for instance. A nice “tutorial overview” of the
catalytic potential and mechanisms for redox non-innocent ligands has been written by
Lyaskovskyy and de Bruin.[173]
O MII
+
M(II)-phenolate complex
O M
2+one-electron
oxidation
formal oxidation state: M(III)
O MIII
M(III)-phenolate
O MII
M(II)-phenoxyl radical
possible experimental oxidation numbers
Scheme 16. Two possible outcomes of a one-electron oxidation of a metal-phenolate complex
(adapted from Shimazaki et al.[170]).
21
Tunability through substituents
Like carbenes, the phenolate ligand offers different sites for modification of its steric and
electronic properties. Usually, substituents are introduced ortho and para to the phenol
oxygen, since these positions are synthetically readily accessible.[152, 167] Upon
complexation, the ortho group is in close proximity to the metal, offering a control element
for the steric environment. Adapting the para group, on the other hand, allows modification
of the electronic properties and solubility of the ligand without affecting the steric demand
of the system.[174]
1.5 Objectives
The objective of this dissertation work was to develop new tripodal mixed NHC/ phenolate
ligands and to explore their coordination chemistry with late transition metals. Mixed
NHC/phenolate ligands with other binding geometries have been reported, such as
bidentate NHC/phenolates,[175-176] vaulted, 4-coordinate bis(NHC-phenolates)[177] or pincer-
type ligands with an OCO binding motif (O = phenolate, C = NHC).[178-181] Most recently,
Bercaw and coworkers published tridentate dianionic mixed ligands with one NHC
moiety.[182] However, to the best of my knowledge, (BIMPNR,R’,R”)– and (MIMPNR,R’,R”)2–
are the first tripodal ligands that combine NHC and phenolate chelating arms.
As stated above, TIMENR iron imido and -nitride complexes were too sterically hindered to
promote group- or atom transfer chemistry. Desirably, side access may be improved by
exchanging one or two carbene arms with phenolates. Although the ortho-phenol groups
can possess substantial steric bulk, as in the case of adamantyl functionalization, the
phenolate’s substituents are further away from the metal center, causing less steric pressure
while still providing sufficient protection, e.g., to prevent bimolecular decomposition
reactions. The new ligands (BIMPNR,R’,R”)– (anion of bis[2-(3-R-imidazol-2-ylidene)ethyl]-
[(3,5-R’,R”-2-hydroxyphenyl)methyl]amine) and (MIMPNR,R’,R”)2– (dianion of mono[2-(3-
R-imidazol-2-ylidene)ethyl]-bis[(3,5-R’,R”-2-hydroxyphenyl)methyl]amine) (Chart 1) can
be regarded as hybrid ligands or cross-overs between the N-anchored tris(carbene) and the
tris(phenolate) ligand systems. Together with the TIMENR and ((R,R’ArO)3N)3– ligands,
they provide a complete ligand series, ranging from tris(carbene) to tris(phenolate), from
22
NN
N
3
R
TIMENR
N
O
3
((R,R'ArO)3N)3-(BIMPNR,R',R'')-
NN
N
2
RO
(MIMPNR,R',R'')2-
NN
NR
O
2
R'
R'' R''
R' R
R'
Chart 1. Series of tripodal N-anchored ligands from tris(carbene) (left) to tris(phenolate) (right).
which the ligand with the most suitable steric and electronic properties for the desired
reactivity may be chosen.
A series of metal complexes with mid- to late transition metals was to be synthesized and
thoroughly characterized. The thesis work focused on (BIMPN)– complexes of Mn, Fe, and
Co as central metals: For these, TIMENR chemistry can serve as comparison to evaluate the
differences and new qualities of the mixed ligands. Furthermore, the capacity of the new
ligand’s complexes for small molecule activation is to be explored, starting with – but not
limited to – the generation of nitrides analogously to Vogel’s work.
23
2 Results and Discussion
About the numbering of the compounds.
Numbers are assigned to the different (BIMPNR,R’,R”)– complexes to refer to them
shorthand. Superscripts may designate different substituents on the NHC and phenolate
groups. If no superscripts are given, the number refers to the complex of the ligand
derivative (BIMPNMes,Ad,Me)–, as this ligand was used more often for several reasons: Its
synthesis, especially in the interimly used tosylation route (section 2.1.4), is cleaner and
easier to handle; more importantly, the complexes of this ligand derivative crystallized
more readily, which is why the majority of the crystal structures explored in the following
chapters stems from this derivative.
When a complex differs only in its counter ion, the counter ion may be added in brackets.
Examples:
[(BIMPNMes,Ad,Me)Fe(Cl)] 1
[(BIMPNMes,tBu,tBu)Fe(Cl)] 1Mes,tBu,tBu
[(BIMPNMes,Ad,Me)Fe]BPh4 1[BPh4]
Complexes in which the (BIMPNR,R’,R”)– ligand has been modified covalently are denoted
by a star *.
A list of all numbered compounds and complexes is given in section 6 (p. 242).
2.1 Synthesis of Mixed Phenolate-Carbene N-Anchored Ligands
2.1.1 Motivation
The effort to create new hybrid ligands that combine NHC and phenolate binding sites in
an N-anchored tripodal ligand framework was driven mainly by two considerations: Firstly,
the reactive cavity of the TIMENR nitrido and imido complexes (and potential oxo
analogues) needed to be opened for side access of organic substrates to enable atom and
24
group transfer reactions. Secondly, the novel ligands would bridge the gap between the
tris(carbene) and the tris(phenolate) ligands employed in the Meyer group laboratories,
thereby providing a continuous ligand series from which to choose a ligand with the most
suited electronic and steric properties for a given task.
For the most part, the efforts described herein are confined to the
bis(carbene)-mono(phenolate) ligand (BIMPNR,R’,R”)– (anion of bis[2-(3-R-imidazol-2-
ylidene)ethyl-(3,5-R’,R”-2-hydroxyphenyl)methyl]amine). The synthesis of the
mono(carbene)-bis(phenolate) ligand (MIMPNR,R’,R”)2– (dianion of mono[2-(3-R-imidazol-
2-ylidene)ethyl-bis(3,5-R’,R”-2-hydroxyphenyl)methyl]amine) was developed in
collaboration with Johannes Hohenberger and elaborated on in his diploma thesis.[183]
2.1.2 The Basic Building Blocks: Imidazoles and Phenols
The ardent interest in NHC chemistry has fostered the development of viable synthetic
routes to substituted imidazoles. A common strategy for NHC formation is the
quaternization of the second nitrogen in the pentacycle, followed by deprotonation of the
positively charged imidazolium to the neutral carbene. Commonly, imidazoles are created
in 1-pot multicomponent syntheses (compare Scheme 18, p. 26) in a modified Debus-
Radziszewski reaction.[184-185] The drawback of this synthesis, especially for 1-aromatic
substituted imidazoles, is the low yield and therefore uneconomical production on higher
scale. Liu et al.[186] found that the yield of 1-aryl-imidazoles can be substantially increased
by a 1-pot-2-step procedure, in which the aromatic aniline is first stirred with the glyoxal at
RT to create in situ an imine. In the second step, the ring closure condensation with
ammonia and formaldehyde is done under reflux. For mesityl-imidazole (1-(2,4,6-
trimethylphenyl)imidazole) and xylyl-imidazole (1-(2,6-dimethylphenyl)imidazole), yields
in excess of 50 % were achieved in the course of this work.
While 2,4-di-tert-butylphenol is a cheap, commercially available starting material, phenols
with the sterically more demanding adamantyl ortho-substituent were synthesized using the
procedures outlined in Scheme 17: The 2-adamantyl-p-cresol (2-adamantyl-4-methyl-
phenol) was obtained by an electrophilic aromatic substitution, catalyzed by concentrated
sulphuric acid at room temperature,[187] 2-adamantyl-4-tert-butylphenol by a slightly
25
modified literature procedure[188-189] in which the pure, solvent-free starting materials are
stirred at elevated temperature.
OH
HO+
(CH2Cl2)
[H2SO4]
OH
OH
Cl+ii) 140°C, 14h
OHi) 100°C, 3h
Scheme 17. Syntheses of 2-adamantyl-p-cresol (top)[187] and 2-adamantyl-4-tert-butylphenol
(bottom).[188-189]
2.1.3 The Original Plan: Mannich-Chlorination-Route
The synthesis of the tris(carbene) TIMENR ligand is well established (Scheme 18) and, for
some substituents, like R = mesityl, 2,6-xylyl, the protonated ligand precursor can be
readily obtained on a multi-gram scale.[65, 71, 190] The tris(phenolate) ligand ((R,R’ArO)3N)3–
requires a very sensitive, acid catalyzed condensation (Scheme 19),[191] but for those
phenols mentioned in the last section it is also a well established ligand in the Meyer
group.[73-75] The syntheses of these C3-symmetric ligands are summarized in Schemes 18
and 19. The first draft for a synthesis of a mixed NHC/phenolate ligand was basically a
combination of these synthesis routes (Scheme 20). In the next section, the synthetic results
following this scheme towards (BIMPNR,R’,R’’)–, as well as the difficulties encountered, are
discussed.
26
R
NH2O
O
NH4
O
H
H
N
N
R
NOH
NCl
AcOH
(MeOH)
reflux
SOCl2(CH2Cl2)
+
(H3TIMENR)Cl3
NN
NR
3
Cl
TIMENR
NN
NR
3
KOtBu
(THF) - KCl
3
3
150 °C,
2d
Scheme 18. Synthesis of TIMENR.
(isopropanol, anhydrous)
[p-TsOH]
OH
R'
R
+ NN
N
NN
OH
R
R'3
(toluene)
NaOMeN
ONa
R
R'3
(R,R'ArOH)3N Na3((R,R'ArO)3N) Scheme 19. Synthesis of the tris(phenolate) ligand ((R,R’ArO)3N)3–.[191],[75]
Mannich reaction
OH
R''
R'
NN
NR
n
OH
R'
R''3-n
n = 2: (H3BIMPNR,R',R'')Cl2
n = 1: (H3MIMPNR,R',R'')Cl
NN
NR
n
OK
R'
R''3-n
n = 2: K(BIMPNR,R',R'')
n = 1: K2(MIMPNR,R',R'')
NOH
n NHO
n
OH
R'
R''3-n
NCl
n
OH
R'
R''3-n
SOCl2
chlorination
SN2
nN
N
R
KOtBu
deprot.
Cl
Scheme 20. Planned synthetic route towards mixed NHC/phenolate ligands (BIMPNR,R’,R’’)– and
(MIMPNR,R’,R’’)2–.
27
First Ligand Batches and Purification Problems
Several batches of ligand precursor (H3BIMPNR,R’,R’’)2+ were synthesized according to
Scheme 21. Derivatives with different substituents were created in the hope that
purification problems (vide infra) could be surmounted by altering solubility and
crystallization properties.
(MeOH / H2O)
OH
R''
R'
NN
NR
2
OH
R'
R''
(H3BIMPNR,R',R'')2+
HNOH
2 NHO
2
OH
R'
R''
NCl
2
OH
R'
R''
SOCl2
(CH2Cl2)
(MeCN)
> 2 eq.
N
N
R
CH2O
c1-R',R'' c2-R',R''
R' = R'' = tBu
R' = Ad, R'' = tBu, Me
R' = R'' = tBu
R' = Ad, R'' = tBu, Me
R' = R'' = tBu
R' = Ad, R'' = Me
R = Mes,
R' = R'' = tBu
R' = Ad, R'' = tBu, Me
R = Xyl,
Scheme 21. First synthetic route towards (BIMPNR,R’,R’’)– and combinations of substituents for
which the synthesis was carried out.
Compounds c1-R’,R” were synthesized in a Mannich type reaction following a slightly
modified procedure by Marinescue et al.[192] Transformation to the chlorinated compounds
c2-R’,R” with thionyl chloride went smoothly. Next, the imidazole moieties were attached
via an SN2 reaction in refluxing acetonitrile, which takes several days for completion. The
reaction can be monitored by 1H NMR spectroscopy, and is deemed complete when signals
of the mono-substituted intermediate – in which only one chloride has been replaced with
an imidazole – have vanished.
Several side products form during the reaction. Test reactions with different solvents or
temperatures gave equal or worse results, or, for lower temperatures, no notable
conversion. While the TIMENR ligand can be purified by salt metathesis and
recrystallization, crystallization has proven to be more intricate for the mixed ligands,
likely due to the change in solubility through the phenol group and/or the ambiphilic
28
behavior caused by the concurrence of the more lipophilic phenol and the hydrophilic
imidazolium groups.
Different solvents, methods (cooling, evaporation, layering technique, slow diffusion) and
counter-anions (PF6–, BPh4
–, ClO4–, –OTs) were tested for crystallization. In one instance, a
fraction of the compound (H3BIMPNXyl,tBu,tBu)2+ could be crystallized via addition of conc.
HCl in EtOAc in the presence of PF6– with subsequent cooling of the mixture. A change in
solubility hints that the crystallized product was most likely protonated at the amino-
function: The crystals were soluble only in alcohols and DMSO, whereas the pro-ligand,
prior to HCl-treatment, was also soluble in chlorinated solvents, THF, and EtOAc. Slightly
different 1H NMR chemical shifts corroborate this assessment, although the N–H proton
signal itself could not be observed due to rapid exchange with the protic solvent and/or
water impurities.3 However, this crystallization process proved unfit for steady ligand
precursor synthesis. Firstly, the amount of HCl the crystals contained did not appear to be
stoichiometric, causing difficulties in the deprotonation step. Efforts to remove the acid
after crystallization were hampered due to the water solubility of the cationic ligand
precursor (in the presence of the Cl– anion), which led to a great reduction in yield when
the ligand was washed with aqueous bases. Stirring the crystallized precursor in organic
solvent with solid K2CO3 resulted in low yields and/or ligand decomposition over time.
Secondly, crystallization of the precursor was difficult to reproduce. The procedure may be
highly sensitive to crystallization conditions, concentrations and/or amount of side
product(s) generated during the SN2 reaction.
NN
OH
Scheme 22. One of the side products formed in the reaction of c2-tBu,tBu with xylyl-imidazole.
3 The NMR solvents used in this case were not dried like the ones used for sensitive compounds under N2
atmosphere.
29
During one attempt at crystallization, one of the side products crystallized neatly and could
therefore be thoroughly analyzed (Scheme 22, for 1H NMR spectra, see p. 170).
Comparison of 1H NMR spectra revealed that this compound was one of the major side
products. It is formed by substitution of the amino group at the reactive benzylic position of
the phenol by an imidazole. While secondary amines are, in general, not favorable leaving
groups, the reaction might be facilitated by intra-molecular hydrogen bonding of the
hydroxyl-proton to the amino-nitrogen in a 6-membered ring conformation (Scheme 23).
This would pre-form the amino-leaving group. Also, hydrogen bonding would decrease the
electron density on the nitrogen, making it more electron-withdrawing and therefore
rendering the neighboring benzylic methylene group even more susceptible to a
nucleophilic attack. The same reaction also leads to mono- and bis(imidazolium)
substituted ethylamines (Scheme 24). (For deliberate synthesis of p2-R, see chapter 2.1.5,
p. 34; the N,N’-benzyl-aryl-imidazolium in Scheme 22 can be synthesized independently
by reacting the chloromethylated phenol p3-R’,R” (chapter 2.1.6, p. 39) with the
imidazole.)
O
R'
R''
H
N
N
R
NR1
R2
- HNR1R2
O
R'
R''
N
R
N
Scheme 23. Reaction pathway leading to cationic side products depicted in Scheme 22 and
Scheme 24.
HNN
NR
2NH
NN
R
Cl
p2-R Scheme 24. Mono- and bis-imidazolium side products of the SN2 reaction between c2-R’,R” and
N-R-imidazole.
30
While most neutral side products as well as excess starting material can be easily removed
by sonication of the crude product with solvents in which the ligand precursor is insoluble
(alkanes, Et2O, sometimes aromatics or EtOAc, depending on substituents and counter
ions), the great difficulty arose from separating the different imidazolium compounds, each
carrying one or two positive charges, from each other. Efforts to purify the ligand precursor
with chromatographic methods, testing different common column materials[193] as well as
reversed phase column chromatography, failed to give clean ligand precursor. While it
might be possible to separate the cationic compounds with special column material,[194] this
approach would be both cost- and time-intensive and thus impractical for multi-gram
batches of the desired compound for the subsequent coordination chemistry.
Protecting Groups
Since the formation of the inseparable, charged side products was rationalized with
hydrogen bonding (Scheme 23), it was reasoned that this side reaction might be suppressed
by attaching a protecting group (PG) on the phenol oxygen. This protecting group must be
cleavable under conditions that do not affect the ligand precursor or its further use. Basic
conditions should be either mild, so as not to deprotonate the imidazolium ring, or suitable
for deprotonation of the entire ligand, so that deprotonation and deprotection may be
carried out in one step prior to metal coordination. Acidic conditions may protonate the
nitrogen anchor, potentially causing the same problems as discussed for
(H3BIMPNXyl,tBu,tBu)2+ crystallized by HCl-addition (vide supra). Hydrogenation might
affect the double bond of the imidazolium rings, although Kariuki and coworkers have used
palladium catalyzed hydrogenation to cleave a benzyl group from bis(imidazolium)amine
compounds in good yields.[195-196]
According to these considerations, TBDMS (tert-butyldimethylsilyl) was chosen as a
suitable PG, since silyl-groups can be cleaved by a fluoride source, which should not affect
the rest of the ligand. However, standard procedures[197] for attaching the group with
TBDMS-Cl, with different bases, varying solvents, and reaction temperatures, showed no
conversion of c2-tBu,tBu according to TLC and 1H NMR.
Subsequently, the benzyl protecting group was tested for its applicability. While cleavage
of the benzyl group through hydrolysis involves the inherent risk of cleaving the benzylic
amine, no disruption of the C–N bond was observed by either Licini or Brown and
31
coworkers during similar reactions towards tris(phenolate) ligands.[198-199] The first
attempts to treat c2-tBu,tBu with benzyl bromide revealed slow conversion and a mixture
of unidentified products (1H NMR).
Next, protection of the phenol as a THP (tetrahydropyranyl) ether was tested. To save
c2-tBu,tBu material, the commercially available phenol was used for the test reaction. This
approach resulted, however, in ortho-substitution of the phenol with the THP group
(Scheme 25). In compound c2-R’,R”, the second ortho position is already blocked by the
nitrogen anchor, so the desired phenol-protection might occur instead. At this point,
however, the protecting group strategy was deferred in favor of other, simpler routes that
showed more promise.
OH
+
O(CH2Cl2)
[p-TsOH]
r.t.
OH
O
Scheme 25. Reaction of 2,4-di-tert-butylphenol with dihydropyran.
Bromination
Introduction of a bromide instead of a chloride leaving group did not effectively alter the
outcome of the SN2 reaction.[200]
Finkelstein Reaction
A different approach to avoid formation of the cationic side products during the SN2
reaction is to improve the quality of the leaving group on the ethyl arms, allowing the
selectivity to be high and/or reaction conditions to be mild enough to prevent substitution at
the benzylic position. An in situ Finkelstein reaction (Scheme 26) substitutes the chloride
with an iodide. For this, chlorinated compound c2-Ad,Me was dissolved in acetonitrile and
five equivalents of solid potassium iodide were added. The mixture was refluxed for three
hours to give intermediate c3-Ad,Me, before a solution of xylyl-imidazole in acetonitrile
was added and the mixture continued to reflux for several days. The reaction with xylyl-
imidazole to (H3BIMPNXyl,Ad,Me)2+, monitored by 1H NMR, occurred faster than without
the Finkelstein step, but the same amount of side products developed. Possibly, side
32
product formation might be suppressed by lowering the temperature after the in situ
Finkelstein reaction has occurred, and/or by adding the imidazole very slowly.
NCl
2
OH
R'
R''
c2-R',R''
KI
(MeCN or acetone)
NI
2
OH
R'
R''
c3-R',R'' Scheme 26. Finkelstein reaction of c2-R’,R” to c32-R’,R”.
It was also attempted to isolate the iodized intermediate. To this end, c2-Ad,tBu was
dissolved in acetone and refluxed with six equivalents of KI. However, subsequent reaction
of c3-Ad,tBu with xylyl-imidazole went exceedingly slow at low temperatures, while the
same product mixture as with the chloride compound c2-Ad,tBu was yielded at elevated
temperatures.
2.1.4 The Tosylation Route
An effective way to create excellent SN2-substrates from alkyl alcohols is to convert them
to alkyl sulfonates. The tosyl group (Ts, para-toluenesulfonyl) provides a particularly good
leaving group, since the emerging tosylate –OTs stabilizes the anionic charge through
delocalization into both the sulfonate and the aromatic ring. Reaction of c1-R’,R” with
TsCl in CH2Cl2 (Scheme 27) yielded the di-tosylated product c4-R’,R” alongside mono-
substituted intermediate c4a-R’,R” and tri-substituted, phenol-tosylated side-product
c4b-R’,R”. Reaction control by TLC can limit the amount of c4b-R’,R” by optimizing the
reaction time and slowly adding more TsCl when it is fully consumed. The amount of
c4b-R’,R” depends on the amount of TsCl that is effectively used, and on the steric
demand of the ortho-substituent R’ on the phenol. c4b-R’,R” formation is small for
R = Ad, larger for R = tBu, and finally for R = Me,[201] no disubstituted product could be
isolated.
Before c4-R’,R” can be used in an SN2 reaction to form (H3BIMPNR,R’,R’’)2+, all mono-
substituted c4a-R’,R” has to be removed, as it will lead to the corresponding mono-
imidazolium-compound and cause the same purification problems as discussed above
33
NHO
2
OH
R'
R''
2 eq. TsCl
(CH2Cl2)
c1-R',R''
0°C - r.t.
NTsO
2
OH
R'
R''
c4-R',R''
NTsO
OH
R'
R''
c4a-R',R''
NTsO
2
OTs
R'
R''
c4b-R',R''
++ Cl
Scheme 27. Tosylation of c1-R’,R”.
(pp. 27ff). The high reactivity renders c4-R’,R” very sensitive, so it has to be handled with
great care. c4-R’,R” has been observed to react even with free chloride anions, if those are
not removed after the tosylation reaction. Column chromatography with silica led to partial
decomposition, and experiments with ethyl acetate as eluent gave a secondary product
which can only have emerged from reaction with this solvent. With the right precautions,
c4-R’,R” can be purified by column chromatography with THF/hexane eluent mixtures on
neutral alumina. Ideally, c4-R’,R” is synthesized right before use, its workup done quickly,
and the compound never heated above room temperature.
While the high reactivity of c4-R’,R” encumbers handling and purification, it finally
yielded the desired results in the synthesis of (H3BIMPNR,R’,R’’)2+: The substitution with
imidazole can now be carried out at room temperature. Clean starting materials provided,
the reaction proceeds smoothly, and none of the plaguing cationic side products are
observed.
The tosylation route was also tried for synthesis of (MIMPNR,R’,R”)2–. It has to be pointed
out that for the bis(phenol) compound, recovery of the desired mono-tosylated product is
even more difficult since two phenols now compete with one alkyl alcohol for tosylation
(see Scheme 28). Therefore, while a first batch of (MIMPNR,R’,R”)2– could be obtained this
way, albeit not entirely pure, this was not deemed a convenient route for the bis(phenolate)
ligand.
34
NHO
OH
R'
R''
TsCl
(CH2Cl2)
0°C - r.t.
++
2
NTsO
OH
R'
R''2
NTsO
OTs
R'
R''2
N
OH
R'
R''
etc.
OTs
R'
R'' OTs
Scheme 28. Tosylation of a bis(phenol)-compound for synthesis of (MIMPNR,R’,R”)2–.
It may be possible to improve the Mannich–SN2 route by picking another, less tricky
leaving group, with a reactivity lying in between the chloride and tosylate leaving groups.
This new leaving group would have to be more stable than the tosylate, but still reactive
enough to maintain the positive effect of mild reaction conditions in the following SN2 step.
Protecting the phenol from tosylation might also be possible. For compound c1-R’,R”,
with its competing OH-groups, this would probably require a multi-PG strategy, adding at
least three extra steps to the overall ligand synthesis (protecting the alkyl alcohol with
selective PG1, protecting the phenol with PG2, removing PG1 to free the alcohol for
tosylation). Alternatively, the PG may be attached to the phenol prior to the Mannich
condensation, which, however, might in turn influence the Mannich reaction since the
electronic and steric properties of the phenol are altered.
2.1.5 The Bis(imidazolium) Routes
While the previous sections describe routes towards mixed NHC/phenol ligands in which
the phenol is attached first to the nitrogen anchor of the ligand, the complementary
“carbene first” routes are explored in the following.
Synthesis of Bis(imidazolium) Salt p2-R
Thionyl chloride chlorinates diethanolamine in near quantitative yield to
di(chloroethyl)ammonium chloride p1×HCl,[202] which reacts with excess aryl-imidazole to
35
the bis(imidazolium) salt p2-R×HCl in a slightly modified literature procedure[203] in
moderate yield (40 %, Scheme 29).
HNOH
2
SOCl2
(CH3Cl)
H2NCl
2
Cl
p1 x HCl
(MeCN)
> 2 eq.
N
N
R
H2NN
NR
2
3 Cl
p2-R x HClR = Mes, Xyl Scheme 29. Chlorination and SN2 reaction leading to bis(imidazolium) salt p2×HCl.
Yields of the protonated salt (H3TIMENR)Cl3 in the TIMENR synthesis (see p. 26) depend
strongly on whether or not the amine has been deprotonated beforehand. However, contrary
to the tertiary tri(chloroethyl)amine employed in the TIMENR synthesis, the deprotonated
secondary amine p1 is prone to polymerization. A portion of HCl-free
di(chloroethyl)amine left to stand at room temperature for a day will turn into a gooey,
yellowish solid, and even storage at low temperatures of the pure substance does not
eliminate this process. This explains why, at the elevated temperatures necessary for the
nucleophilic substitution at the chloride, only the imidazole starting material and an
insoluble residue could be retrieved when the di(chloroethyl)amine was deprotonated
beforehand.
This issue is even more pressing for the mono(halidoethyl)amine necessary for
(MIMPNR,R’,R”)2– synthesis, namely mono(bromodiethyl)amine-hydrobromide. The
imidazole alone is basic enough to sufficiently deprotonate this primary amine for
polymerization.[183, 204] The amine-hydrobromide is therefore added in small portions to a
solution of a four-fold excess of imidazole, to keep local concentration of the amine low
and suppress polymer formation.[183] A corresponding procedure might also increase yields
of p2-R.
Attaching the Phenol
Several procedures were tried for coupling the phenol to the anchoring unit of the
bis(imidazolium) salt p2. Mannich type reaction between phenol and p2×HCl with either
paraformaldehyde or an aqueous solution of formaldehyde showed no conversion
36
(Scheme 30). The protonated amine may not be able to condense with the formaldehyde to
form the imino-intermediate, however, addition of base (NEt3 or pyridine) in slight excess
did not further the reaction, nor did carbonate salts (NaHCO3 or K2CO3) that were stirred in
the reaction mixture.
OH
+CH2O
37% aq.+
H2NN
NXyl
2
3 Cl
p2-Xyl x HCl
(MeOH)
reflux, days
(MeOH)
reflux, days
1 eq. base(NEt3 or pyridine)
OH
+(CH2O)n
(solid)
H2NN
NXyl
2
3 Cl
p2-Xyl x HCl
(MeOH)
reflux, hours
(MeOH)
reflux, several days
monitored by 1H NMR Scheme 30. Attempts at attaching the phenol to the nitrogen anchor of p2-R×HCl by Mannich
condensation (examples).
Reductive amination has been used successfully to synthesize tris(phenolates)[198-199] and
asymmetric N-anchored ligands that combine for instance pyridyl or pyrazolyl groups with
phenolates.[205-206] Accordingly, bis(imidazolium) salt p2-Xyl×HCl and 2,4-di-tert-butyl-
salicyl aldehyde were treated with sodium cyanoborohydride (NaBH3(CN)) as reducing
agent in a MeOH / THF mixture. The protic solvent activates and enhances the reactivity of
the borohydride, while THF (or another non-chlorinated, suitable solvent) is necessary to
dissolve the salicyl aldehyde (Scheme 31). However, low conversions of p2-Xyl×HCl
were accompanied by reduction of the bulk of the salicyl aldehyde to the benzyl alcohol. In
order to prevent direct reduction of the aldehyde, different reaction conditions and
procedures were tested in an effort to enforce imine formation between the two starting
materials prior to addition of the reducing agent: p2-R×HCl and the aldehyde were stirred
37
for prolonged periods of time; water-removing agents, such as 3 Å molecular sieves, were
added to favorably shift the equilibrium and bases (NEt3 or pyridine) were added to
deprotonate the amine. None of these procedures substantially enhanced the reaction
outcome.
OH
+H2N
NN
Xyl
2
3 Cl
p2-Xyl x HCl
O i) 1h to days*, RT, **
ii) NaBH3(CN)OH OH
(MeOH / THF)
(H3BIMPNXyl,tBu,tBu)2+
+
(< 10%)
Scheme 31. Attempts at attaching the phenol to the nitrogen anchor of p2-R×HCl by reductive
amination (examples), *stirring time range was varied, **optional addition of molecular sieves
(3 Å) or base (Et3, pyridine).
Both Mannich type reaction and reductive amination may have failed because the amine
group in p2-R×HCl was protonated and its nucleophilic attack on a carbonyl to form the
imide/iminium ion-intermediates hindered. Since in situ addition of bases to the reaction
mixture was unsuccessful, it was tried to isolate the hydrochloride-free bis(imidazolium)
salt p2-R as starting material for these reactions.
p2-R×HCl was stirred with NaHCO3 in CH2Cl2, the idea being that while the hydro-
chloride adduct is only very moderately soluble in CH2Cl2, the favorable lattice energy of
NaCl and CO2-evolution would favorably shift the equilibrium, and free, CH2Cl2-soluble
p2-R could conveniently be filtered off the solids. This approach was only limitedly
successful: conversion was slow, and the extract contained side products apparently from
decomposition of the bis(imidazolium) amine.
When the bis(imidazolium) salt was stirred in aqueous NaHCO3 solution, any attempts to
extract it with organic solvents were unsuccessful due to the high solubility of both free
p2-R and its hydrochloride in water. When, however, a suitable counterion is added to the
mixture, it is possible to transfer p2-R into the organic phase. That way, it was possible to
extract HCl-free p2-R[PF6] with CH2Cl2 in 82 % yield (for R = xylyl) after the chloride
was exchanged by hexafluorophosphate (Scheme 32).
38
H2NN
NR
2
3 Cl
p2-R x HCl
i) NaHCO3 aq.
ii) > 2eq. NaPF6
ii) extraction with CH2Cl2 HNN
NR
2
2 PF6
p2-R[PF6] Scheme 32. Deprotonation of the N-anchor of bis(imidazolium) salt p2-R.
HCl-free p2-Xyl[PF6] was converted to (H3BIMPNXyl,tBu,tBu)2+ by reductive amination in
40 % yield (by 1H NMR), with 60 % direct reduction of the aldehyde to the benzylic alcohol
(see Scheme 33). Several possibilities remain to optimize the reaction: As mentioned, protic
solvents are necessary to activate the borohydride. Accordingly, the solvent (or mixture of
solvents) may be changed to adjust the reactivity (and selectivity) of the reducing agent.
Also, the exact pH value might be crucial. While HCl apparently hinders the reaction, acetic
acid has been used as catalyst for a similar reductive amination,[205-206] suggesting that a
milder acid may as well be able to further formation of the iminium intermediate. Finally,
other hydrogenation agents should be tested. In particular, sodium triacetoxyborohydride
NaBH(OAc)3 has been appraised as “a superior, convenient, and effective reducing agent for
reductive amination reactions”, which also “eliminate[s] the risk of residual cyanide, not only
in the product but also in the workup waste stream”.[207] It has been used in synthesis of
tris(phenolates)[198-199] and asymmetric glucose-containing ligands.[208]
The application of p2-R[PF6] in the Mannich reaction has not been tested, nor have
possible alternative reaction conditions and procedures for reductive amination been
exhaustively explored, since another, viable route towards the mixed NHC/phenolate
ligands emerged at the time.
OH
+HN
NN
Xyl
2
2 PF6
p2-Xyl[PF6]
O i) 1h, RT
ii) NaBH3(CN)OH OH
(MeOH / THF)
(H3BIMPNXyl,tBu,tBu)2+
+
(40%)
Scheme 33. Reaction of p2-Xyl[PF6] with di-tert-butyl salicyl aldehyde and NaBH3(CN).
39
2.1.6 Final Synthetic Route for (BIMPNR,R’,R”)– and (MIMPNR,R’,R”)2–:
Chloromethylation of the Phenol and SN2 Reaction with
Bis(imidazolium) Salt p2-R
Chloromethylphenol p3-R’,R” was generated via a Blanc reaction in a slightly modified
literature procedure (Scheme 34),[183, 209] in which gaseous HCl was substituted by HCl
conc. This change improves ease and safety of handling, while the reaction time increases
significantly (1 d for p3-tBu,tBu, > 2 d for p3-Ad,Me). Vigorous stirring is crucial for
conversion in the now biphasic reaction mixture. Great care must be taken during workup:
In the Blanc reaction, part of the formaldehyde can react to bis(chloromethyl)ether in
hydrochloric solutions. This substance is highly irritant as it decomposes to formaldehyde
and hydrochloric acid in wet air (e.g. inside the lungs), and its alkylating properties render
it one of the most potent synthetic carcinogens.[210-211]
OH
R''
R'HCl conc.,
CH2O
(toluene)
OH
R''
R'Cl
p3-R',R"
Scheme 34. Chloromethylation of the phenol to yield p3-R’,R”.
In the presence of base, p2-R×HCl reacts with p3-R’,R” to (H3BIMPNR,R’,R”)2+
(Scheme 35). The chloromethylated phenol can self-react intermolecularly to a spiro
compound (via HCl-elimination followed by Diels-Alder cycloaddition), therefore the
reaction vessel is first charged with the bis(imidazolium) salt and p3-R’,R” is added very
slowly and in small excess.[183, 212] The reaction can be tested for full conversion of
p2-R×HCl by 1H NMR spectroscopy, by evaporating a small sample of the reaction
mixture and dissolving the residue in DMSO-d6, since remaining bis(imidazolium) salt is
hardly removable from the ligand precursor (see pp. 27ff). After completion, the reaction
mixture is evaporated to dryness, the residue taken up in methanol, heated to reflux, and the
ligand precursor is precipitated by slow addition of a warm solution of > 2 eq. NaBPh4 in
methanol. The abundant, voluminous, white precipitate can be filtered off, further purified
through different filtration and washing steps if necessary (see experimental section), to
40
successfully give analytically pure (H3BIMPNR,R’,R”)(BPh4)2 in high to near quantitative
yields.
i) NEt3, reflux
ii)
(CH2Cl2)
OH
R''
R'Cl
H2NN
NR
2
3 Cl
p2-R x HCl
NN
NR
2
OH
R'
R''
(H3BIMPNR,R',R'')(BPh4)2(MeOH)
NaBPh4
2 BPh4
p3-R',R"
Scheme 35. Synthesis of the pro-ligand (H3BIMPNR,R’,R”)(BPh4)2 via SN2 reaction between
p2-R×HCl and p3-R’,R”.
Deprotonation of the Pro-Ligand
Several bases were tested for the deprotonation of the ligand precursor to the free, mono-
anionic ligand (BIMPNR,R’,R”)–. Commonly and conveniently, potassium tertiary-butoxide
(KOtBu) is used in THF, which provides the free ligand in near quantitative yields
(Scheme 36). After removing the solvent from the reaction mixture, the ligand can be
extracted from the residue with diethyl ether. The drawback of this method is that emerging
tBuOH perseveringly sticks to the product, turning the carbene into a glassy, sticky residue,
and sometimes causing difficulties during complexation to a metal center (see
chapter 2.2.1). If necessary, the tertiary alcohol can be removed by lyophilization from
benzene (several cycles are necessary; the tenacity of the alcohol to stay in the product may
be attributed to H-bonding to the phenolate).
KOtBuN
NN
R
2
OH
R'
R''
(H3BIMPNR,R',R'')(BPh4)2
2 BPh4
NN
NR
2
OK
R'
R''
K(BIMPNR,R',R'')
(THF)
Scheme 36. Deprotonation of the pro-ligand with KOtBu to the potassium salt of the free NHC/
phenolate ligand.
41
Benzyl potassium (BnK), prepared by metalation of toluene with Schlosser’s base
(n-butyllithium and KOtBu), also provides free ligand. The ligand solution or slurry in THF
has to be cooled before slow addition of BnK in small portions to ensure selectivity of the
deprotonation. Otherwise, the reaction proceeds markedly less cleanly than with KOtBu.
The other drawback in comparison with KOtBu is that BnK is not commercially available
but has to be freshly prepared since, due to its extremely high reactivity, it can only
limitedly be stored. The advantages on BnK lie in the bright orange color of the reagent,
which disappears upon reaction, conveniently pointing out when enough base has been
added (as soon as the orange color remains), and, above that, in the easy removal of the
generated toluene.
Deprotonation with sodium methanolate NaOMe leads to free carbene as well. The 1H NMR spectrum of the sodium salt Na(BIMPNR,R’,R”) differs from the spectrum of the
potassium salt, in that several chemical shifts are different and some signals broader. This
hints at some interaction of the ligand with the different alkali cations, which is
corroborated by the crystal structure of the potassium salt (vide infra). While the lower
boiling point/higher vapor pressure of methanol should render it more easily removable
than tBuOH, the H-bonding effects (vide supra) seem to play a role just as well.
Furthermore, it has been shown that the carbene can insert into the C–O bond of
methanol,[213] which may explain some of the impurities in the sodium salt 1H NMR
spectrum. The tertiary alcohol seems to be less prone to undergo this reaction. In summary,
methanol also stays tenaciously in the ligand substance, and leads to less pure ligand
batches, and therefore cannot be recommended.
Other bases that generate volatile side products include sodium hydride and alkali salts of
hexamethyldisilazane (HMDS). NaH generates H2, which would evaporate even during
deprotonation. Sadly, reaction of NaH with (H3BIMPNR,R’,R”)2+ (with –OTs of –BPh4
counter ion) gave low isolated yields of ligand with low purity. NaH was tested with and
without the use of DMSO as catalyst.[113] With NaHMDS, the reaction mixture turned into
a very sticky slurry, from which no free ligand could be extracted. Cooling of all reactants
before mixing and use of different solvents gave no better results. Therefore, on the whole,
KOtBu has been established as most efficient base, despite the persistent tert-butyl alcohol.
42
2.1.7 Crystal Structures of a (BIMPNR,R’,R”)– Ligand and its Protonated
Precursor
Colorless needles of (H3BIMPNMes,Ad,Me)(OTs)2 suitable for single crystal X-ray analysis
were obtained by slow evaporation of a solution of the compound in a CH2Cl2 /
ethyl acetate mixture (Figure 1).
Figure 1. Crystals of (H3BIMPNMes,Ad,Me)(OTs)2: Colorless needles grown in heaps upon slow
evaporation of a solution in a CH2Cl2 / ethyl acetate mixture, left and right: in daylight and under
the microscope.
The crystal structure of (H3BIMPNMes,Ad,Me)(OTs)2 is characterized by stacks of cations and
anions along the crystallographic b-axis (Figure 2). Within these stacks each imidazolium
cation forms two intermolecular C–H···O hydrogen bonds to one of the tosylate anions that
points towards the cavity of the ligand. The remaining O–H donor function of the
imidazolium cation forms an intramolecular O–H···N hydrogen bond with the anchor
nitrogen atom N1. The second tosylate anion forms separate stacks and does not show
pronounced interactions with the cations.
Crystals of the ligand were grown by cooling an ethereal solution of the ligand’s potassium
salt to –35 °C. The molecular structures of the cation of (H3BIMPNMes,Ad,Me)(OTs)2 and of
[K2(BIMPNMes,Ad,Me)2(C6H6)] are shown in Figures 3 and 4; selected geometrical structure
parameters are given in Table 22 (p. 224).
43
Figure 2. Packing in the crystals of (H3BIMPNMes,Ad,Me)(OTs)2 · 0.2 EtOAc along the
crystallographic b-axis.
In the crystal structure of [K2(BIMPNMes,Ad,Me)2(C6H6)] two potassium cations are
coordinated by two ligand molecules each. The phenolate oxygen atoms of both ligands
link the two potassium cations in a µ-η1-η1-fashion. The coordination sphere around K1 is
completed by the two anchoring N atoms and two carbene C atoms from two different
ligand molecules, thus forming a heavily distorted octahedron. In contrast to this, the
coordination around the second potassium K2 can be described as distorted tetrahedral.
This cation is coordinated by the two oxygen donors, one carbene donor atom, and a
solvent benzene molecule acting as a fourth donor in an η6 binding mode. The η6 benzene
turned out to be rotationally disordered with two preferred orientations.
The N–C bond distances and N–C–N angles within the protonated imidazole and
deprotonated imidazole-2-ylidene all agree well with literature values.[142] Upon
deprotonation, these bond lengths increase by an average 0.04 Å and the angles decrease by
about 7°.
44
Figure 3. Molecular structure of the (H3BIMPNMes,Ad,Me)2+ cation in crystals of
(H3BIMPNMes,Ad,Me)(OTs)2 · 0.2 EtOAc (50 % probability ellipsoids, the tosylate counter ions, co-
crystallized solvent molecules, and – with the exception of the phenol and imidazole H atoms –
hydrogen atoms are omitted for clarity.)
Figure 4. Molecular structure of [K2(BIMPNMes,Ad,Me)2(C6H6)] in crystals of
[K2(BIMPNMes,Ad,Me)2(C6H6)] · 3 Et2O. (50 % probability ellipsoids, co-crystallized solvent
molecules and hydrogen atoms are omitted for clarity.)
45
2.1.8 Summary of Ligand Synthesis and Outlook
The synthetic routes towards mixed NHC/phenolates that have been described in this
chapter can roughly be divided into two categories: On the one hand, the “phenol first”
routes, in which the phenol is attached to the N-anchor by Mannich condensation with
diethanolamine, followed by introduction of a leaving group to the compound that can be
replaced with the imidazole. On the other hand are the complementary “carbene first”
paths, in which the phenol is attached secondly to the bis(imidazolium-ethyl)ammonium
(p2-R, for (BIMPNR,R’,R”)–) or mono(imidazolium-ethyl)ammonium (for
(MIMPNR,R’,R”)2-).
The “phenol-first” routes often led to mixtures of cationic imidazolium compounds,
generated by nucleophilic attack of an imidazole at the benzylic position between N-anchor
and phenol. The different ionic imidazolium compounds are difficult to separate and purify.
This issue was resolved when tosylate was introduced as leaving group, which allows mild
reaction conditions in the SN2 step and therefore selective substitution on the ethylene
chains only. The high reactivity of the tosyl-group renders the intermediate compound
c4-R’,R” metastable and places high demands on its handling and purification. For steady
production of larger quantities of ligand it was therefore considered cumbersome and
further pathways were tested. However, the tosylation approach does open up a viable
route, especially for ligand derivatives with precious imidazoles, since the imidazole is
attached at a late stage in near quantitative yields, whereas the synthesis of the
(imidazolium-ethyl)ammonium compounds entails high losses of imidazole due to only
moderate yields.
The “carbene-first” route, on the other hand, avoids large-scale column chromatography
and metastable intermediates. Mannich condensation between the phenol and p2-R×HCl
has not yet yielded ligand precursor, and reductive amination reactions still have to be
thoroughly optimized to avoid direct reduction of a considerable part of the salicyl
aldehyde to the benzyl phenol. Superior outcomes were achieved by SN2 reaction of
p2-R×HCl with the chloromethylated phenol p3-R’,R”.
46
OH
R'
R''
H3-nNX
n
x HX
nN
N
R
H3-nNN
x HX
NR
n
X
(X = Cl, Br)
OH
R'
R''
Cl
NN
NR
n
OH
R'
R''3-n
(toluene)
HCl conc.
n = 2: (H3BIMPNR,R',R'')2+
n = 1: (H3MIMPNR,R',R'')+
NN
NR
n
OK
R'
R''3-n
n = 2: K(BIMPNR,R',R'')
n = 1: K2(MIMPNR,R',R'')
KOtBu
CH2O
Scheme 37. Final synthetic route for K(BIMPNR,R’,R’’) and K2(MIMPNR,R’,R’’).
Scheme 37 summarizes the total synthesis of the new mixed NHC/phenolate ligands. This
now established route gives good overall yields of pure ligand, finally providing an
excellent synthetic pathway for steady production of (H3BIMPNR,R’,R”)2+ and
(H3BIMPNR,R’,R”)+ with alkyl and aryl substituents on a multi-gram scale.
Outlook: Further Derivatives and Water Soluble Ligands.
The next chapters expand on complex synthesis and characterization with the bis(carbene)-
mono(phenolate) ligand (BIMPNR,R’,R’’)–, mainly with the derivatives (BIMPNR,tBu,tBu)– (R
= xylyl, mesityl) and (BIMPNMes,Ad,Me)–. In the future, a wide variety of substituents may
be combined on the ligand to fine-tune the complexes’ reactivity. In the meantime, the
tosylation approach is applied by Eva Zolnhofer towards potentially water-soluble ligands
synthesized with precious imidazoles bearing hydrophilic groups.
47
2.2 Iron Complexes of (BIMPNR,R’,R”)–
2.2.1 Iron(II) Complexes
Under inert atmosphere, treatment of (BIMPNMes,Ad,Me)– with one equivalent of anhydrous
ferrous chloride in diethyl ether and a small amount of pyridine at room temperature yields
the four-coordinate iron(II) complex [(BIMPNMes,Ad,Me)Fe(Cl)] (1) as yellow powder in
about 80 % yield (see Scheme 38). Pyridine is necessary unless all tBuOH generated during
ligand deprotonation has been removed completely (which is a tedious process, as
described in section 2.1.6). Otherwise, the tertiary alcohol in combination with the iron salt
creates acidic conditions capable of re-protonating the carbenes. Presumably, the alcohol’s
O-H-bond is activated by coordination of the oxygen to the iron ion, which acts as Lewis
acid (Scheme 39).
NN
NMes
2
OK
Ad
(py / Et2O)
FeCl2
- KCl
(py / Et2O)
FeBr2
- KBr
N
NC
NC
N
FeIIO
N
Br
N
NC
NC
N
FeIIO
N
BPh4N
NC
NC
N
FeIIO
N
NN
N
NaN3
(DMF)
-NaCl NaBPh4
- NaCl
(THF)
N
NC
NC
N
FeIIO
N
Cl
Scheme 38. Synthesis of various (BIMPNMes,Ad,Me))– iron(II) complexes.
48
FeII
N
NR
OH
Scheme 39. Re-protonation of the ligand in the presence of tBuOH.
Unlike the corresponding [(TIMENR)Fe(II)] complexes, the (BIMPNR,R’,R’’)– complexes do
not precipitate from neat pyridine, and therefore the mixture of pyridine and ether was
chosen, which facilitates isolation of the complex while still guaranteeing basic conditions.
Complex 1 is well soluble in pyridine, acetonitrile, DMSO and THF, less soluble in
chlorinated solvents, and insoluble in less polar solvents like diethyl ether and
hydrocarbons. While 1 is stable in dichloromethane, chloroform is capable of oxidizing the
Fe(II) complex, especially under elevated temperatures (vide infra). It is very sensitive
towards O2 and/or H2O, even in the solid state, and turns grayish brown upon exposure to
air. 1Mes,tBu,tBu and 1Xyl,tBu,tBu, which can be prepared analogously, show similar behavior,
only with somewhat higher solubilities in THF and chlorinated solvents. This could be
observed for all following complexes: In general, equivalent complexes of the derivatives
with tBu-substituents on the phenol are more readily soluble than their adamantyl-cresol
counterparts, and sometimes in less polar solvents (i.e. in a wider range of solvents). The
tBu-derivatives also tend to form gooey and sticky masses instead of powders or crystalline
material, and in crystallization setups, very often they form oily droplets rather than
crystals.
The corresponding complex [(TIMENMes)Fe(Cl)]Cl is insoluble in THF and MeCN, and as
a rule of thumb, for a given metal, less polar solvents are necessary to dissolve complexes
of (BIMPNR,R’,R’’)– than their TIMENR counterparts (R = Mes, Xyl) with the same NHC
substituents and counterion(s).
The paramagnetic 1H NMR spectra of 1 (MeCN-d3, CDCl3 or DMSO-d6) feature signals
ranging from 55 to –10 ppm, some of which are strongly broadened and, in part,
overlapping with each other, thus preventing unequivocal integration and signal
assignment. Regardless, the spectra – and especially two moderately sharp peaks around
49
50 ppm with relative intensities of 3 : 1 – are very characteristic and allow the
identification and determination of the purity of the complex. Traces of H2O lead to
additional signals in this region, hinting at the formation of an aqua or hydroxo complex,
since the oxidized iron(III) complex is considered NMR silent (vide infra).
While the solid state molecular structure, derived by X-ray crystallography, is of C1
symmetry (vide infra) with two complexes related by an inversion center in the unit cell,
the 1H NMR spectra in polar solvents suggest a dynamic behavior in solution, resulting in
Cs symmetry on the NMR time scale. The structure of the complex in solution also appears
to vary depending on the solvent’s ability for metal coordination and solvation of the
chloride counter ion, as is indicated by the markedly different signal distribution (and
higher overall number of signals) in the 1H NMR spectrum in THF-d8 compared to MeCN-
1H NMR spectrum of a sample of 2 photolyzed for 22 h in THF-d8 (ca. 17 mg in
0.8 mL). The inset depicts part of the spectrum with an overlay of the spectrum of the starting
material 2 (orange line), of which all signals have disappeared. No further product signals were
observed from 170 to –130 ppm.
While the 1H NMR (Figure 18) looks promisingly neat, giving reason to hope that the
sample contains only one (or one major) well-defined species, the only crystal ever
obtained from a photolyzed solution of 2 was again a crystal of the azido complex. Most
notably, this crystal grew from a fully photolyzed NMR sample, which exhibited a
spectrum that hardly showed any starting material peaks any more. Clearly, the azide
complex crystallizes far more readily than any potential irradiation product.
6 See also section 2.7.2 for the insertion reaction of the [(BIMPNMes,Ad,Me)Co(N3)] irradiation product.
67
Figure 19. Solutions of 2 in THF (left cuvette) and MeCN (right cuvette) before (A) and after (B)
27 h of irradiation with a mercury vapor lamp.
Further experiments may encompass monitoring the reaction by UV/Vis (Figure 19). The
safest way to ensure that it is in fact the azide ligand that is photolyzed may be by
photolysis of a KBr pellet of 2, monitored with IR spectroscopy to see the azide stretches
disappear.7 To ensure that the complex does not simply decompose by oxygen or air
moisture diffusing into the pellet during the time of irradiation, another KBr pellet may be
kept in the dark and measured as reference sample. However, in the end, with a
paramagnetic product, the most conclusive form of identification may remain single crystal
analysis.
7 For this experiment, see also section 2.7.2.
68
2.3 Manganese Complexes of (BIMPNR,R’,R”)–
Treatment of (BIMPNMes,Ad,Me)– with one equivalent of anhydrous manganese chloride in
an ether / pyridine mixture (15:1) at room temperature yields the four-coordinate
manganese(II) complex [(BIMPNMes,Ad,Me)Mn(Cl)] (3) as a white powder in 55 % yield
(Scheme 38). This solvent mixture was chosen for the same reasons as discussed for the
iron(II) complexes (see p. 47).
NN
NMes
2
OK
Ad
N
NC
NC
N
MnIIO
N
(py / Et2O)
MnCl2
- KCl
N
NC
NC
N
MnIIO
N
NN
N
NaN3
(MeCN)
-NaCl
Cl
Scheme 40. Synthesis of the (BIMPNMes,Ad,Me))– manganese(II) complexes [(BIMPNMes,Ad,Me)MnCl]
(3) and [(BIMPNMes,Ad,Me)Mn(N3)] (4).
The IR spectrum of 3 is very similar to that of iron complex 1. Salt metathesis from
chloride to azide affords colorless [(BIMPNMes,Ad,Me)Mn(N3)] (4), which is easily identified
by the intense and characteristic νas(N3) IR vibrational bands (in KBr: 2077 and 2056 cm-1,
Figure 20).8
8 For a discussion on the number of νas(N3) vibrational bands (two where one is expected), see the respective
section for cobalt azide complex 7 on p. 83.
69
Figure 20. IR spectrum of 4 (KBr pellet).
The UV/Vis-spectra of both complexes feature bands with maxima at 257 and 309 nm of
similar intensity (ε = 11300 and 5750 Μ−1 cm–1 for 3, 14900 and 7400 Μ−1 cm–1 for 4),
which are tentatively assigned to a π–π* transition in the phenolate and a charge-transfer
transition, respectively.[220] As expected, no d-d transitions can be observed as they are both
Laporte and spin forbidden.
Both manganese(II) complexes are NMR-inactive, which is to be expected from
manganese(II) d5 high spin systems. Variable temperature SQUID magnetization
measurement (2 – 300 K) confirms a high spin ground state for 3 (Figure 21). At room
temperature, 3 possesses a magnetic moment, µeff, of 5.82 µB (S = 5/2 ground state) that
decreases negligibly with decreasing temperature down to 10 K. From this temperature on
it drops to a value of 5.13 µB at 2 K. The simulation of the measurement returns a g-value
of 1.97 and a zero-field splitting |D| of 0.426 cm-1. Simulation of a variable field
measurement (Figure 22) confirms the zero-field splitting’s magnitude and gives it a
negative sign.
70
Figure 21. Variable temperature SQUID magnetization data (1 T) of [(BIMPNMes,Ad,Me)MnCl] (3).
Magnetic moment (µeff) plotted versus temperature (T). Data were corrected for diamagnetism, and
reproducibility was verified by measuring multiple independently synthesized samples. Parameters
are discussed in the text and listed in Table 16.
Figure 22. Temperature-dependent SQUID magnetization data of [(BIMPNMes,Ad,Me)MnCl] (3) at
variable field strength; magnetic moment (µeff) plotted versus temperature (T). Data were corrected
for diamagnetism. Parameters are discussed in the text and listed in Table 16.
71
Crystal Structure of 3
Crystals of 3 suitable for X-ray crystal structure determination were obtained as colorless
prisms by slow diethyl ether diffusion into a pyridine solution of the complex. The
molecular structure of this manganese complex (Figure 23 and Table 1) exhibits C1
symmetry. The manganese(II) central ion is coordinated by two NHC carbenes, the
phenolate oxygen, and the chloride in a distorted trigonal pyramidal geometry with the
manganese located above the plane of the three donor atoms O1, C3, and C8 (towards Cl1)
as is indicated by the corresponding out-of-plane shift doop of 0.362(2) Å. Notably, a non-
coordinated pyridine molecule is situated between the adamantyl and one of the mesityl
substituents. One of its C–H bonds is pointing towards the Mn-Cl bond at a distance of ca.
3 Å (see Figure 24), which does not suggest any bonding interaction, but the pyridine
effectively pushes apart two of the ligand arms increasing the angle between them to (Ccarb.–
M–O) = 130.69(8). This observation establishes the intended steric flexibility of the
(BIMPNR,R’,R”)– ligand to provide side access for possible substrates to the reactive center.
Figure 23. Molecular structure of [(BIMPNMes,Ad,Me)MnCl] (3) in crystals of
[(BIMPNMes,Ad,Me)MnCl] · 0.5 pyridine (50 % probability ellipsoids). Co-crystallized solvents and
hydrogen atoms are omitted for clarity. Selected bond distances and angles are collocated in
Table 1.
72
Table 3. Selected bond distances [Å] and bond angles [°] for [(BIMPNMes,Ad,Me)MnIICl] · 0.5
pyridine (3 · 0.5 py); see Chart 2 for atom labeling. Further values are listed in Tables 23 and 24.
Bond / Angle 3 · 0.5 py
M···Nanchor 2.696
M–Laxial 2.4221(7)
M–Ccarb. 2.227(3)
2.230(3)
M–O 2.001(2)
Nanchor–M– Laxial 170.25
Ccarb.–M–Laxial 104.79(7)
102.65(7)
O–M–Laxial 92.45(5)
Ccarb.–M–C’carb. 109.02(9)
Ccarb.–M–O 130.69(8)
111.61(8)
NNHC1–Ccarb.–NNHC2 103.1(2)
103.5(2)
doop 0.362(2)
Figure 24. Space-filling representations of the molecular structure of 3 in crystals of
[(BIMPNMes,Ad,Me)MnIICl] · 0.5 pyridine; side view (left and middle) and top view (right) down the
Mn–Cl axis, with and without the pyridine molecule situated between two ligand arms.
73
Outlook
The work on (BIMPNR,R’,R”)– manganese complexes was taken up by Eva Zolnhofer in her
master’s thesis, where she finished the full characterization of the azide complex 4,
including X-ray crystal structure analysis, EPR, and SQUID magnetization measurements.
These data have been published in a collaborative journal publication.[221]
The first photolysis experiments on the azide 4 were performed in the course of the
master’s thesis, with interesting preliminary results. Also, the anion exchange product
[(BIMPNMes,Ad,Me)Mn]BPh4 was fully characterized, and may serve as a more reactive
entrance molecule (compared to 3) for small molecule activation reactions.
74
2.4 Electrochemistry
To probe the redox-activity of the iron complexes [(BIMPNMes,Ad,Me)Fe]Cl (1) and
[(BIMPNMes,Ad,Me)Fe(N3)]Cl (2), as well as that of the cobalt complexes
[(BIMPNMes,Ad,Me)Co]X (X = Cl: 6, X = PF6: 6[PF6], and X = N3: 7) (see section 2.6.1),
electrochemical measurements (cyclic, linear sweep, and square wave voltammetry) were
performed. Linear sweep measurements were used to ascertain the identity of the processes
as oxidation- or reduction events. Square wave voltammetry was applied to discern
potentials that lie closely together. The results are summarized in Table 4 and explained in
the following. Cyclovoltammograms not shown here can be found in the experimental
section.
Table 4. Potentials of reversible redox processes for complexes 1, 2, 6, 6[PF6], and 7 from cyclic
voltammetry in acetonitrile; wave separations at scan rate 0.2 V/s. Potentials are given vs. Fc/Fc+.
compound
reversible
reduction
MI/M
II
E1/2 [V]
wave
separation
∆E [V]
reversible
oxidation
(MII
/MIII
and others)
E1/2 [V]
wave
separation
∆E [V]
[(BIMPNMes,Ad,Me)Fe]Cl 1 - - -0.31 0.6
[(BIMPNMes,Ad,Me)Fe(N3)] 2 - - -0.33 0.66
[(BIMPNMes,Ad,Me)Co]Cl 6 -2.0 0.07 -0.36
0.34
0.60
1.1
0.08
0.18
[(BIMPNMes,Ad,Me)Co]PF6
6[PF6] -2.0 0.07 -0.23
0.60
0.85
0.12
[(BIMPNMes,Ad,Me)Co(N3)] 7 -2.0 0.08 -0.43
0.32
0.60
1.1
0.07
0.08
75
As the chlorido and azido anions are not coordinated to the FeII and CoII complexes of
(BIMPNMes,Ad,Me)– in acetonitrile solution, the (BIMPNMes,Ad,Me)– complexes of the same
metal unsurprisingly give rise to very similar voltammograms in this solvent, regardless of
their “axial ligand” (i.e. counter ion in solution). Each voltammogram features a reversible
redox process assignable to the MII/MIII oxidation, with an unusually large peak separation
that can be explained by a rearrangement of the coordination sphere upon electron transfer
(Figure 25). This rearrangement could stem from coordination of either the counter ion (Cl–
or N3–), coordination of solvent molecule(s), or even a change in the coordination of the
(BIMPNMes,Ad,Me)– ligand.
Figure 25. Cyclic voltammogram of [(BIMPNMes,Ad,Me)Fe]Cl (1) in MeCN: FeII/FeIII redox-wave at
different scan rates.
Cobalt complexes 6, 6[PF6], and 7 additionally show a metal-centered reversible reduction
(CoII/CoI) at –2.0 V, with a peak separation of 0.07 V (Figure 26). While the expected peak
separation ∆E for a reversible redox event is 59 mV in water (and other protic solvents), it
is usually larger in aprotic solvents, and thus ∆E = 0.07 V is common for a reversible
process in acetonitrile. Furthermore, the cobalt complexes’ cyclic voltammograms show
either one or two additional reversible oxidations depending on the counter ion (Figure 26
(left side) and Figure 27): 6 and 7 exhibit an oxidation around E1/2 = 0.3 V, again with a
small peak separation ∆E = 0.07 V (underlined values in Table 4), and all three have an
76
oxidation at E1/2 = 0.6 V. The latter may be assigned to a reversible ligand oxidation. Since
the cyclic voltammogram of the (protonated) ligand (H3BIMPNMes,Ad,Me)(BPh4)2 (see
Figure 65 p. 165)9 shows one irreversible oxidation at 0.72 V, this is in agreement with the
observation that the oxidation potentials of metal coordinated phenolates are generally
lower than those of free phenols.[170]
Figure 26. Voltammograms of 6[PF6]: Cyclic (green line) and linear sweep (maroon line)
voltammograms (left) and isolated CoI/CoII redox-wave at different scan rates (right).
Figure 27. Voltammograms of 6, 6[PF6], and 7 in MeCN.
9 For (H3BIMPNXyl,tBu,tBu)Cl2, the irreversible oxidation lies at 0.78 V (Figure 66 p. 167).
77
The exact mechanism to which the oxidation wave at E1/2 = 0.3 V of 6 and 7 can be
attributed has not been elucidated. What may be ruled out is that these are simply the
oxidations of the free Cl– and N3– anions: The N3
–/N3 potential has been reported to be
0.92 V vs. Fc/Fc+ in acetonitrile,[222] or 1.32 V vs. NHE in H2O,[223] and Cl– has a higher
potential still.
As a working hypothesis, the following tentative explanation may be given: Upon
oxidation, the Cl– and N3– (which had been solvatized counter ions) coordinate to the now
more positively charged metal center. Through mechanistics not yet understood, this
facilitates the oxidation of the counter ions, which now occur at 0.32 V. This oxidation is
reversible. The counter ion may stay coordinated until the cobalt center is reduced to CoII
again. This might also explain why the CoII/CoIII peak separation is somewhat smaller for
6[PF6] than for the other two: The coordinated counterion hinders the reduction to MII,
which then occurs at a lower potential, whereas the bulky PF6– cannot coordinate. (The
smaller wave separation for 6[PF6] is why the effectively observed E1/2 is less negative,
even though the oxidation half-wave is found at the same potential). However, with this
explanation it is on the other hand surprising that the oxidation at 0.6 V is identical in all
three voltammograms; it would mean that the oxidation of the phenolate ligand is not
influenced by a coordinated and oxidized chloride or azide.
78
2.5 Diamagnetic Complexes of (BIMPNR,R’,R”)–
While the isolated manganese(II) complexes of (BIMPNR,R’,R”)– are NMR silent, the
iron(II) complexes give paramagnetic 1H NMR spectra with paramagnetically shifted and
broadened signals that are diagnostic for the identity and purity of the respective
compounds, but not assignable to individual protons. It was therefore of interest to
synthesize at least one diamagnetic compound of the new tripodal ligand in order to receive
interpretable (1H and 13C) NMR spectra, which may also answer questions about the
complexes’ symmetry in solution. Furthermore, for future electrochemical studies, it may
be highly favorable to have diamagnetic reference complexes with a non-redox-active
metal center at hand. Provided the redox-waves are comparable, this may allow an
unambiguous assignment of metal-based and ligand-based redox processes in the
voltammograms of the redox-active complexes.
Accordingly, several metals were tested in closed-shell oxidation states that are expected to
lead to diamagnetic complexes with centers that are redox-inactive on a broad potential
range, including the main group elements Ga(III) and In(III) (Scheme 41) and transition
metal Ni.
GaCl3 A(BIMPNR,R',R")
A = Na, K
+
R = Xyl, R',R" = tBu
R = Mes, R' = Ad, R" = Me
(benzene
or Et2O)
" [(BIMPNR,R',R")M]2+ "
(H3BIMPNR,R',R")2+
+
M = Ga, In
(THF or
py/Et2O)in different ratios
InCl3 A(BIMPNR,R',R")
A = Na, K
+
R = Mes, R' = Ad, R" = Me
Scheme 41. Synthetic routes that were tried towards isolable gallium and indium complexes of
(BIMPNR,R’,R”)–.
Gallium trichloride salt was reacted with the potassium salt of the ligand. In 1H NMR
spectra of the crude product, new peaks are found in the diamagnetic region with a
distribution and coupling patterns that belong to neither a protonated form nor an alkali salt
of the ligand. Comparison with the spectra of the Zn complex (see below) provides further
evidence that these stem from ligand coordinated to the diamagnetic gallium metal center.
79
However, the samples also contained the protonated form of the ligand. Due to similar
solubilities of both the protonated ligand (H3BIMPNR,R’,R’’)2+ and the putative
[(BIMPNR,R’,R’’)Ga]2+, isolation of a pure Ga(III) complex has not been achieved.
The ligand may have been re-protonated, either by impurities in the M(III) trichloride salts,
which may contain traces of HCl, or by acidic conditions through interaction of tBuOH and
the metal salt, i.e., by the same phenomenon as in the iron complex synthesis (see p. 47). In
the first case, recrystallization of the metal salt may remove these impurities; in the latter,
protonation might be counteracted by the same practice as with iron, addition of pyridine to
the reaction mixture to ensure basic conditions.
With InCl3, complexation was carried out in neat THF and in a mixture of py/Et2O.
However, in THF, the same problem with re-protonated ligand arises. After synthesis with
pyridine, strangely enough, a small amount of re-protonated ligand was observed as well.
In addition, even after work-up in different solvents, the pyridine had not been removed
from the crude product. While it is conceivable that one pyridine molecule may be
coordinated to In, the pyridine signals’ integrated intensities strongly exceed 1 equivalent
of pyridine compared to the supposed complex signals. The pyridine signals also
superimpose several of the product signals in the aromatic region. However, NMR spectra
of the crude product still suggest complex formation: The ethylene-groups on the ligand
arms split into four signals, just like in the Zn complexes’ spectra (see below); for isolation
of the complex, however, the synthesis and/or work-up procedure has to be revisited.
Exposing the samples of “[(BIMPNR,R’,R’’)M]2+“ (M = Ga or In) to air did not change the 1H NMR spectra, which suggests that the metal complexes are, as expected, not susceptible
to oxidation by air.
Suggestions for future work on the group 13 compounds include the use of (more) pyridine
and/or a stronger base to avoid ligand protonation, as well as a 1-pot approach like for
[(BIMPNMes,Ad,Me)Zn]OTs (5[OTs], see below) to achieve direct coordination of the freshly
deprotonated ligand molecules.
With Ni(COD)2 (COD = cycloocta-1,4-diene), the free ligand reacts in THF to a light
orange compound. The field desorption mass spectrum (FD-MS) features a prominent peak
at 690 amu, which corresponds to a singly positively charged “[(BIMPNXyl,tBu,tBu)Ni]“.
80
However, further work on this compound has been deferred, since the work with Zn(II)
quickly lead to more promising results.
Experiments with anhydrous zinc chloride lead to readily isolable complexes (see
Scheme 42). Colorless [(BIMPNMes,Ad,Me)Zn]OTs (5[OTs]) was generated in a 1-pot
reaction between the three white solids (H3BIMPNMes,Ad,Me)(OTs)2, NaOMe, and ZnCl2 in
THF. Its 1H NMR spectrum substantiates that, on the NMR time scale, the complex has Cs
symmetry in solution: In the aliphatic region, four signals from 4.5 to 3.0 ppm couple with
each other (see Figure 28), as confirmed by H,H–COSY. These are assignable to the two
times four protons of the two ethylene bridges between N-anchor and NHC-rings, meaning
that the CH2 protons that were equivalent in the uncoordinated ligand are now
diastereotopic. The CH2 protons between anchor and phenolate are still equivalent
(enantiotopic), resulting in one singlet at 3.63 ppm. Furthermore, the two “sides” of one
mesitylene ring (o- and m-positions of the mesityl–substituent) are no longer equivalent,
and the adamantyl substituents give more intricate coupling patterns than in the free ligand.
Hence, neither substituent seems to rotate freely around its bond to the NHC- or phenolate-
ring (on the NMR time scale). The 31 carbon signals found in the 13C NMR spectrum of
5[OTs] (Figure 29) confirm Cs symmetry.
When the ligand’s potassium salt K(BIMPNMes,Ad,Me) is used instead of the 1-pot pathway,
the synthesis with ZnCl2 leads to the chloride complex [(BIMPNMes,Ad,Me)ZnCl] (5). Its 1H NMR spectrum is equivalent to the one to 5[OTs], naturally lacking the tosylate signals,
with only minor changes in chemical shifts and coupling constants.
(H3BIMPNR,R',R")OTs2
(THF)
[(BIMPNMes,Ad,Me)Zn]OTsNaOMe
ZnCl2
K(BIMPNR,R',R")
(THF)
[(BIMPNMes,Ad,Me)Zn(Cl)]
ZnCl2
Scheme 42. Synthesis of [(BIMPNMes,Ad,Me)Zn]+ complexes 5[OTs] (top row) and 5 (bottom row).
[(TIMENXyl)Co(N3)]BPh4 · MeCN. See Chart 2 for atom labeling and Tables 26 and 27 for more
parameters.
Bond / Angle 7 · 2 THF
· 0.5 C6H6
[(TIMENXyl)Co(N3)]BPh4
· MeCN
M···Nanchor 2.659 3.213
M–Laxial 2.039(2) 1.938(2)
M–Ccarb. 2.081(2)
2.060(2)
2.052(2)
2.049(2)
2.017(2)
M–O 1.931(2) -
Nα–Nβ 1.188(3) 1.161(3)
Nβ–Nγ 1.164(3) 1.169(3)
Nanchor–M– Laxial 168.7 174.27
Ccarb.–M–Laxial 102.23(8)
105.10(8)
102.17(8)
101.85(8)
110.48(8)
O–M–Laxial 89.93(7) -
M–Nα–Nβ 128.2(2) 166.3(2)
Nα–Nβ–Nγ 177.1(2) 178.3(2)
Ccarb.–M–C’carb. 106.95(8) 118.86(8)
111.91(8)
110.52(8)
Ccarb.–M–O 134.03(7)
112.31(7)
-
NNHC1–Ccarb.–NNHC2 102.8(2)
103.1(2)
103.2(2)
103.6(2)
103.3(2)
doop 0.297(2) 0.520
94
2.6.2 Co(I) Complexes and their Reactivity
One goal in the course of this work was to reproduce the synthesis of terminal cobalt imido
complexes demonstrated by Hu et al. with the TIMENR ligand (R = Xyl, Mes,
Scheme 44).[70] Hu’s imido complexes did not undergo the desired transfer reaction of the
imido group onto organic substrates, but instead the imido group inserted into the Co–
carbene bond. By virtue of a more accessible cobalt imido bond through the introduction of
the phenolate, the (BIMPNR,R’,R’’)– system was expected to provide a platform for
aziridination reactions.
As entrance molecule to the imido chemistry, a Co(I) complex was to be synthesized.
When different (BIMPNR,R’,R’’)– derivatives were reacted with the lime green Co(I)
precursor CoCl(PPh3)3 in benzene or THF, brown solutions formed, and 1H NMR spectra
N
N
N
CN
C
N
CoI NC
N
R
RR
+
R' N3
N
N
N
CN
C
N
CoIII NC
N
R
RR
N
+
_ 35°C, _ N2
R'
R = H, Me
R' = Me, MeO
for all reactions:
solvent, RT
N
N
NC
N
C
N
CoI NC
N
R
R
N
R
R'
N
N
NC
NC
N
CoII NC
N
R
R
N
R
R'
+ "Co(0) species"
+ organic residue
fast
disproportionation
insertion
2+
Scheme 44. Synthesis of terminal cobalt(III)-imido complexes by Hu et al. and subsequent
insertion into the cobalt-carbene bond.[70]
95
confirmed the formation of paramagnetic species. Product purification, however, proved
troublesome, particularly the removal of the nascent triphenyl phosphane. While the
TIMENR cobalt(I) complexes [(TIMENR)CoI]+ synthesized by Hu (R = Xyl, Mes)[70] and
Kropp (R = Mes, iPr)[225] precipitate from benzene during synthesis and can easily be
filtered off and washed, the (BIMPNR,R’,R’’)– Co(I) complexes are soluble in ether, and, for
R’ = R’’ = tBu, even in pentane. The presence of the triphenyl phosphane seems to increase
the complexes’ solubilities.
Product contamination with phosphane is not only a hindrance towards isolation and
characterization of a clean Co(I) product, but also highly adversary to the synthesis of the
Co-imido complexes: Phosphanes react quickly with organic azides (Staudinger reaction),
and a corresponding organic byproduct was observed by 1H NMR spectroscopy. Only with
an excess of organic azide, and after conversion of the entire phosphane, could the
remaining azide attack the cobalt(I)-complex, and the difficulty of product separation
remained. Therefore, another route to the cobalt(I)-complexes was sought.
When green cobalt(II) complex [(BIMPNMes,Ad,Me)Co]X (X = Cl, PF6) is suspended in
benzene and stirred with potassium graphite (KC8), a brown slurry forms. Filtration and
lyophilization yields a brown powder that gives the same paramagnetic 1H NMR signals
that were observed in the reaction between (BIMPNMes,Ad,Me)– and CoCl(PPh3)3.
The complex is soluble in both polar and aromatic solvents, very soluble in ethers and
even, to some extent, soluble in aliphatic hydrocarbons. Elemental analyses so far have
consistently given low C,H,N-values for the cobalt-(I)-complex. This may be due to
residual KCl that stayed in the material even after several cycles of extraction with ether or
benzene and evaporation/ lyophilization, through interactions e.g. of the potassium cation
with the complex phenolates (vide supra). Nevertheless, temperature dependent SQUID
magnetization measurements confirm successful reduction: The reproducibly obtained
value of µeff(RT) = 3.1 µB (see Figure 39) is far lower than the values of the d7 S = 3/2
cobalt(II) complexes 6[PF6] and 7. It is still higher than the spin-only value for a d8 S = 1
system (µs.o. = 2.83 µB), implying contribution of spin-orbit coupling. The identity of the
product as [(BIMPNMes,Ad,Me)CoI] (8) is furthermore established by its crystal structure and
by its follow-up chemistry (vide infra).
96
Figure 39. Temperature-dependent SQUID magnetization data (1 T) for five independently
synthesized samples of [(BIMPNMes,Ad,Me)CoI] (8). Magnetic moment (µeff) plotted versus
temperature (T). Data were corrected for diamagnetism.
Complex 8 is highly reactive towards oxygen and other oxidants, such as iodosylbenzene
(PhIO), which re-oxidize monovalent 8 to the cobalt(II) oxidation state, as can be observed
both by the color change back to green and by 1H NMR spectroscopy.11 It is however fairly
stable towards elevated temperatures, as even prolonged refluxing in acetonitrile did not
change either color or 1H NMR spectrum of the compound.
Crystal Structure of [(BIMPNMes,Ad,Me)CoI] (8)
Suitable crystals for X-ray crystal structure analysis were grown by cooling a saturated
ethereal solution to –35 °C. The crystal structure contains three crystallographically
independent molecules in the asymmetric unit, all of which are structurally and metrically
very similar. Table 8 lists parameters for one of these molecules, parameters for all three
are given in Tables 26 and 27.
11 Reaction of 8 with PhIO gave single crystals and the crystal structure of [(BIMPNMes,Ad,Me)CoII]I (6[I]
· 0.5 MeCN · 0.5 THF). Due to its high similarity with the structures of 6 and 6[PF6], the molecular structure
of 6[I] is not further discussed.
97
The metal center is coordinated solely by the tripodal ligand, and indeed no other
molecules are found in the crystal lattice. Compared to the cobalt(II) complexes 6, 6[PF6],
and 6Xyl,tBu,tBu[PF6], the Co–O and Co–Nanchor bonds in 8 are longer on average by 0.07 Å
and 0.08 Å, respectively, as is expected from σ-bonding ligands when the metal center
becomes more electron rich. Contrarily, the two Co–carbene bonds are shorter by about
0.1 Å, which is readily explained by a strengthened metal-to-ligand π-backbond. This has
also been observed for FeII-FeI complexes[226] and NiI-Ni0 complexes[66] of the TIMENR
ligand. This is accompanied by NCHN–CCarb. bonds that are about 0.02 Å longer in 8 than in
the cobalt(II) complexes.
Figure 40. Molecular structure of the complex [(BIMPNMes,Ad,Me)CoI] (8) (50 % probability
ellipsoids). Hydrogen atoms are omitted for clarity. Selected bond distances and angles are listed in
Table 8.
98
Table 8. Selected bond distances [Å], bond angles [°], and doop [Å] with e.s.d.’s in parentheses for
[(BIMPNMes,Ad,Me)CoII] (8). See Chart 2 for atom labeling.
Bond / Distance 8 Angle 8
M···Nanchor 2.194(2) Nanchor–M– Laxial -
M–Ccarb. 1.923(2)
1.937(2)
Ccarb.–M–C’carb. 111.60(7)
M–O 1.949(2) Ccarb.–M–O 122.58(6)
123.78(6)
doop –0.160 NNHC1–Ccarb.–NNHC2 102.1(2)
101.7(2)
2.6.3 Reactions of [(BIMPNMes,Ad,Me)CoI] (8) with Organic Azides
The reactivity the cobalt(I) complex 8 towards several organic azides was tested. Blank test
were conducted to ensure that conversion of the azide can only be attributed to complex 8:
ArN3 (Ar = Mes, Ph) was added to CoCl2 in THF or benzene and monitored over time.
Within 1 week, no conversion was observed. The test was repeated in the presence of PF6–,
to the same result.
When trityl azide (Ph3CN3) was reacted with 8 in THF, the brown solution immediately
turned blue; in benzene, blue solid immediately precipitated. The product was identified as
the azide complex 7, and in the reaction liquid remained the trityl radical or Gomberg’s
dimer, respectively (see Scheme 46, p. 109). The trityl substituent may be too bulky for
formation of an imido species within the reactive cavity of 8, and consequently it releases
its azide group instead.
When red-brown 8 is reacted with less bulky aryl azides (RN3 with R = Ph, Mes) in either
THF or benzene, the solution turns dark green. 1H NMR reveals total conversion of 8 after
addition of 1 eq of the azide; additional azide remains unreacted in the solution, so no
follow-up reactions occur with the azide. After work-up, the reactions yield green solids in
91 % and 35 % yield for PhN3 and MesN3, respectively.
99
Reaction with PhN3
In the reaction with PhN3, effervescence is observed. The products’ signals are distributed
around the chemical shift region of diamagnetic compounds, albeit no coupling patterns
can be distinguished and some signals are situated in the negative region. Elemental
analysis of the isolated green solid is in agreement with the sum formula for an imido
complex “[(BIMPNMes,Ad,Me)Co=NPh]”. However, variable temperature SQUID
magnetization measurement of this compound reveals a RT magnetic moment µeff(RT) =
4.10 µB close to the values for the (BIMPNR,R’,R”)– cobalt(II) complexes (Figure 41),
whereas the cobalt-imido complex would be expected to possess a cobalt(III) center, like
the diamagnetic d6 cobalt(III) imido species of Hu and Meyer. In the insertion reaction
observed by Hu et.al., the initial insertion product would be cobalt(I), whereas all crystal
structures of the insertion products revealed cobalt(II) centers. Presumably, the cobalt(I)
species undergoes a disproportionation reaction to the cobalt(II) species and elemental
Co(0) (see Scheme 44).
Figure 41. Temperature-dependent SQUID magnetization data (1 T) of the green product from the
reaction between [(BIMPNMes,Ad,Me)CoI] (8) and PhN3. Magnetic moment (µeff) plotted versus
temperature (T). Data were corrected for diamagnetism.
Growth of single crystals of the green compound has not been achieved; however, a few
violet crystals suitable for X-ray crystallographic structure analysis were obtained by
100
diffusion of pentane into a benzene solution of the green compound. The molecular
structure derived from these violet crystals revealed a dinuclear cobalt-complex
[bis-µ-OH{Co(BIMPNMes,Ad,Me*NPh)}2] (9*, Figure 42 and Table 9). The dinuclear
complex is situated on a crystallographic inversion center of the space group (P1̄ ) and
exhibits Ci molecular symmetry. Each cobalt center is coordinated in distorted tetrahedral
geometry by one NHC carbene atom and three oxygens, namely the phenolate and the two
bridging OH groups. The other carbene carbon atom has reacted with the phenyl azide to
form an imine. Most plausibly, this reaction proceeded in the same fashion as demonstrated
for the TIMENR species, i.e. imido complex formation followed by insertion of the imido-
group into the metal-carbene bond. The imino-NHC group thus formed has detached itself
from the metal center. The bond length suggest a fragment of the form Cphenyl–N=CCarbene,
i.e., with stronger double bond character between the nitrogen and the carbene carbon atom.
Figure 42. Molecular structure of the complex [bis-µ-OH{Co(BIMPNMes,Ad,Me*NPh)}2] (9*) in
crystals of [bis-µ-OH{Co(BIMPNMes,Ad,Me*NPh)}2] · C6H6 (50 % probability ellipsoids).
Co-crystallized solvent and hydrogen atoms except for those of the bridging OH groups are omitted
for clarity. Selected bond distances and angles are summarized in Table 9.
101
Table 9. Selected bond distances [Å], angles [°], and doop [Å] with e.s.d.’s in parentheses for
[bis-µ-OH{Co(BIMPNMes,Ad,Me*NPh)}2] · C6H6 (9* · C6H6). See Figure 42 for atom labeling.12
Bond / Angle # Bond / Angle 9* · C6H6
M···Nanchor Co1···N1 2.510
Co1–O2
Co1–O2A
(= Co-µ-O)
1.995(2)
2.052(2)
M–Ccarb. Co1–C3
Co1–C8
2.066(2)
-
M–O Co1–O1 1.938(2)
N7–C8 12 1.297(3)
N7–C47Phenyl 1.399(3)
C8–N7–C47Phenyl 126.0(2)
O2–Co1–O2A 81.15(7)
Co1–O2–Co1A 98.85(7)
O1–Co1–O2
O1–Co1–O2A
116.89(7)
98.84(7)
Ccarb.–M–O C3–Co1–O1
C8–Co1–O1
110.35(8)
-
C3–Co1–O2
C3–Co1–O2A
130.83(8)
104.89(8) #corresponding bond or angle in complexes of (BIMPNMes,Ad,Me)– with unaltered tripodal ligand.
Assuming that the bridging OH-groups stem from water, the most likely source seems
moisture in the glovebox atmosphere. This would be in keeping with the extended time
necessary of crystallization for this compound13 – if the water were in the solvent, it would
be available at once, whereas atmospheric moisture may diffuse into the solution over time
– and also the very small quantity that was obtained of this violet compound. On the other
hand, it might be that the green compound reacts only slowly with the available moisture.
In any case, this dinuclear compound is, obviously, vastly more prone to crystallization
than the green compound.
12 No nitrogen atom in this structure was labeled N6. 13 The crystal structure was reproduced, and in both instances crystal growth took several months.
102
The release of the imino-NHC arm from the metal center is probably induced by the
coordination of the oxygens. It is not clear whether the insertion reaction into the Co–carbene
bond was also induced by formation of the new Co–O bond(s); however, the SQUID and
NMR data on the green compound suggest that the insertion reaction took place beforehand.
Reaction with MesN3
In the reaction of 8 with MesN3, no gas evolution is observed. Judging by the abundance of
signals, and also by their integrals compared to those of the Co(I) complex before the
reaction (versus solvent peak), more than one species is observed within minutes of azide
addition. Also, the spectrum of samples kept at RT changes within the course of the next
days. Therefore, no meaningful assignment of the spectra has been possible.
Green, block-shaped single crystals suitable for X-ray single crystal analysis were obtained
by layering a benzene solution with pentane and hexane. The crystals contained complex
[(BIMPNMes,Ad,Me*N3Mes)CoII] (10*, Figure 43), in which the cobalt center is coordinated
in distorted tetrahedral geometry by the nitrogen anchor, the phenolate oxygen, one NHC
carbene carbon atom, and an abnormal carbene: the second NHC ring has rotated around its
Figure 43. Molecular structure of 10* in crystals of [(BIMPNMes,Ad,Me*N3Mes)CoII] · 0.461 C6H6 ·
0.539 n-pentane (30 % probability ellipsoids). Co-crystallized solvent molecules and hydrogen
atoms are omitted for clarity. Selected bond distances and angles are summarized in Table 10.
C3
C47
103
own axis, and is now coordinated to the cobalt center with the atom CCarb2, the carbon atom
of the double bond which is nearer the nitrogen anchor. The mesityl azide unit is still intact
and its terminal nitrogen atom bound to the former (normal) carbene carbon atom. The zig-
zag shaped N3-unit’s bond lengths suggest the following mode of binding: C3Carb.=N6–
N7=N8–C47Phenol, i.e., alternating single and double bonds.
Table 10. Selected bond distances [Å], angles [°], and doop [Å] with e.s.d.’s in parentheses for
#corresponding bond or angle in complexes of (BIMPNMes,Ad,Me) with unaltered tripodal ligand. ## with C4 (=CCarb2) instead of C3 (=CCarb.) as third atom to define the plane.
104
The molecular structure of 10* explains why no effervescence is observed during this
reaction, since contrary to expectations, no dinitrogen was released from MesN3. The
reason for the very different behavior of 8 towards MesN3 vs. PhN3 may be found in the
azides’ different steric demand. In general, an organic azide may coordinate to a metal
center either linearly through its terminal nitrogen or in bent form through the substituted
nitrogen (Scheme 45). Nitrogen can only be released in the latter case, and the ortho-
methyl-groups of MesN3 may be too sterically demanding for this form of coordination
within the reactive cavity of [(BIMPNMes,Ad,Me)Co]. This explanation is in agreement with
the observation that neither azide employed by Hu et al.[70] for the synthesis of TIMENR
cobalt imido complexes – para-tolyl- and para-anisyl-azide[227] – bears ortho-substituents.
Co
N
N
N
Co
NN
N
Scheme 45. Linear (left) and bent (right) coordination modes of the aryl azide.
Outlook: Suggestions for Further Experiments
Insights gained from the two crystal structures of [bis-µ-OH{Co(BIMPNMes,Ad,Me*NPh)}2]
(9*) and [(BIMPNMes,Ad,Me*N3Mes)CoII] (10*), combined with mechanistic considerations,
have revealed that only organic azides that are sterically less demanding than MesN3 may
form a Co=NR imido group within the reactive cavity of the ligand scaffold
(BIMPNMes,Ad,Me)–. Promising candidates for other aryl azides to be tested are those
employed by Hu[70, 227] and further aryl azides with no ortho-substituents. Interesting would
also be the reaction of 8 with alkyl azides such as AdN3 and iPrN3. Alternatively,
(BIMPNMes,Ad,Me)– derivatives with less bulky substituents, especially on the phenolate’s
ortho-position (R’), may be tested.
105
Addition of styrene to 8 followed by addition of MesN3, at RT or at low temperatures, has
been tested with the intention to transfer an imido group onto the organic substrate, before
it became known that this azide does not form an imido species with the (BIMPNMes,Ad,Me)–
complexes. This experiment may be repeated with PhN3, with either stoichiometric
amounts of styrene or another alkene, or in neat cyclohexane as both solvent and reactant.
106
2.7 Towards a Cobalt Nitride
2.7.1 Introduction
To the best of my knowledge, no mononuclear, terminal cobalt nitride has been
characterized and published. Isolated examples of group 9 nitrido complexes include
iridium[228-229] complexes that feature tridentate, meridional donor ligands, resulting in an
overall square planar geometry. In 2010, Chirik and coworkers reported the formation of
cobalt amide and imine complexes after photolysis of a square planar cobalt azides.[230]
This was rationalized by C–H activation of the ligand by a putative intermediate cobalt
nitrido complex, and the mechanism was elucidated by experiments with deuterated ligand.
However, the putative cobalt nitride could neither be isolated nor was any spectroscopic
evidence provided.
As the ligand field splitting (see Scheme 7, p. 11) suggests, and as the history of the iron
nitrido complexes has shown, it is plausible that the cobalt nitrido group may be well-
stabilized within a tripodal ligand field.
2.7.2 Photolysis of [(BIMPNR,R’,R”)Co(N3)] (7)
As stated in section 2.6.1, in solutions of polar solvents, divalent cobalt(II) complexes of
(BIMPNMes,Ad,Me)– exist in monocationic form. Therefore, it is not surprising that attempts
to photolyze (or thermolyze) the azide complex 7 in MeCN, chloroform, or DMSO lead to
no conversion, even upon prolonged irradiation: The solutions’ color and the 1H NMR
spectra of irradiated samples remain unchanged, since the azide is not coordinated to the
cobalt center. Complex 7 is insoluble in less polar solvents like benzene, therefore THF
remains as the only solvent for photolysis experiments. In the following, if not otherwise
noted, all photolysis experiments are carried out with a Heraeus 150 W mercury vapor
lamp.
107
Bulk and 1H NMR Experiments
Thermolysis of 7 is not feasible, since the complex is stable in THF even upon prolonged
heating, and naturally also in more polar solvents. When THF solutions of the sapphire
blue complex 7 are irradiated, they turn dark green (Figure 44). Upon irradiation of 1H NMR samples of 7 in THF-d8, the azide complexes’ signals vanish, while no distinctive
new signals emerge.
Figure 44. A much diluted solution of 7 (ca. 5 mg) in THF (ca. 8 mL), from left to right: before,
after 1 h, and after 2 h of irradiation.
When an NMR sample was irradiated at 309 nm (± 10 nm) with a LOT 1000 W Xe-OF arc
lamp, some small, paramagnetic peaks formed from 42 to –12 ppm. However, the relative
intensities (determined using the solvent signal height as internal standard for integration)
were about 1/10th of the starting complex at most, meaning that even if those signals can be
attributed to a well-defined photolysis product, the yield of this product would be very
small.
Neither from NMR experiments nor from several bulk photolyses at either RT or with
water or methanol cooling (ca. 10 °C and –10 °C, respectively) was it possible to isolate or
crystallize a new compound.
Trapping experiments
Since the direct product of irradiation remained elusive, several reagents were tested in
NMR and/or vial (20 mL) scale photolyses to trap a possible nitride intermediate. The
electronic structure of the nitride decides which sort of reagent is suitable for trapping
reactions: Closed-shell nitrides, which can be nucleophilic or electrophilic in nature, may
108
be captured by Lewis acids or bases respectively,[230-231] while radicals may be more
suitable for open-shell nitrides.[229] Furthermore, H-atom donors may stabilize a nitride in
the form of an imide or amide, or, if the nitride undergoes insertion into the metal-carbene
bond as was observed by Vogel,[226] may further the reaction towards this product.
Table 11 summarizes the results from photolysis experiments on the NMR scale and/or in
vials with a diverse range of trapping reagents, which are explored in depths in the
following.
Table 11. Trapping reagents used in photolysis experiments with 7.
1H NMR observations upon photolysis:
reagent#
immediate
reaction conversion
of reagent further comments
B(C6F5)3 yes - no further change
TMS-I yes - (immediate precipitation from THF)
nBu3P no no (see Figure 44)
Ph3P no no (see Figure 44)
(Ph3C)2 no no (Gomberg’s dimer, see Scheme 46)
styrene no no -
TEMPO no yes some small new peaks
tBuNC no yes some small new peaks
tBu3-phenol no yes well-defined paramagnetic product
# see text for abbreviations.
Photolysis in the presence of tris(n-butyl)- or triphenyl phosphane (nBu3P and Ph3P),
Gomberg’s dimer ((Ph3C)2, Scheme 46)[232], 14 or styrene lead to no conversion of the
respective reagent despite full conversion of 7. These reagents did not seem to have an
influence on the outcome of the photolysis, even though the presence of the phosphanes
leads to brownish colors of the photolyzed solutions (see Figure 48).
14 As shown in section 2.6.2, reaction of 8 with trityl-azide (Ph3C-N3) gives 7. When done in THF, this leads
to a solution containing 7 and the trityl radical / Gomberg’s dimer, ready for photolysis.
109
C2 CC
Scheme 46. Solution equilibrium of trityl radicals and their dimerization product (Gomberg’s
Dimer).
Figure 45. Solutions of 7 in THF after 4 h of irradiation without reagent (left) and with Ph3P
(middle: 1 eq, right: 16 eq).
When tris(pentafluorophenyl)borane (B(C6F5)3) is added to a THF solution of 7, the color
immediately changes to a brownish green, the 1H NMR spectrum becomes practically
identical to that of 6[PF6], and irradiation leads to no further change. Most likely, the
borane seizes the N3– molecule and forms an anionic Lewis acid / base adduct, leaving the
cobalt(II) complex cation behind (Scheme 47, top row). On addition of trimethylsilane-
iodide (TMS-I) to the THF solution of 7, a light green solid precipitates, which re-dissolves
in MeCN-d3 to give a spectrum akin to 6. Presumably, the TMS-I leads to ion exchange,
leaving the cobalt(II) complex with an iodide counter ion, [(BIMPNMes,Ad,Me)Co]I, which is
insoluble in THF just like its chloride congener (Scheme 47, bottom row).
110
N
NC
NC
N
CoIIO
N
NN
N
+
N
NC
NC
N
CoIIO
N
I
N
NC
NC
N
CoIIO
N
blue
(THF)
(THF)
B(B6F5)3
TMS - I
B(B6F5)3N3
+ TMS - N3
brownish green
light green precipitate Scheme 47. Reactions of 7 with B(C6F5)3 and TMS-I.
If 7 is irradiated in THF solution in the presence of tert-butyl-isocyanide (tBuNC) or
tetramethylpiperidinoxyl (TEMPO), the reagent is consumed while the azide diminishes.
With tBuNC, some new signals grow in the diamagnetic region, though their broadness and
lack of observable coupling declare that they belong to a paramagnetic species. Their
integrals are smaller compared to the original azide signals. The reaction towards the
compound(s) generating these signals seems either very unselective, or most of the
product’s signals are not observable. Some new signals mainly in the diamagnetic region
emerge also in the presence of TEMPO, again with comparably small integrals, but less
broadened.
2,4,6-Tris-tert-butylphenol (tBu3-phenol) can serve as H-atom donor, since the
corresponding phenoxyl radial is fairly stable.[233-234] Upon irradiation of 7 in the presence
of tBu3-phenol, the reagent’s signals are without exception shifted slightly up-field and
broadened, indicating the generation of the phenoxyl radical. Undoubtedly, a new
paramagnetic species emerges (Figure 46). Its signals grow in the range from 53 to
111
-10 ppm and their integrals are comparable to those of the starting material, i.e. the
reaction towards this compound appears to be fairly selective, promising high isolable
yields. In addition, it is noteworthy that the irradiated solutions become emerald green,
whereas solutions that lack the trapping reagent often turn a brownish or grayish hue of
green. The spectrum of the tBu3-phenol “trapping product” is tentatively evaluated and
integrated in the experimental section (p. 208).
Several vial- and larger-scale photolyses were run in an effort to isolate this compound.
The work-up has been optimized by Eva Zolnhofer, who has further characterized the
denoted as 11* · THF · 0.16 H2O). See Figure 47 for atom labeling.
Bond / Angle# Bond / Angle 11*
· THF · 0.16 H2O
M···Nanchor Co1···N1 2.098(2)
M–Ccarb. Co1–C3
Co1–C8
-
1.989(3)
Co1–N6 1.979(2)
M–O Co1–O1 1.887(2)
N6–C3 1.323(3)
C3–N6–Co1 130.4(2)
Ccarb.–M–C’carb. N6–Co1–C8 109.9(2)
Ccarb.–M–O C3–Co1–O1
C8–Co1–O1
-
118.5(1)
N6–Co–O1 118.91(9)
NNHC1–Ccarb.–
NNHC2
N2–C3–N3
N4–C8–N5
106.5(2)
103.7(2)
doop doop
## –0.404
#corresponding bond or angle in complexes of (BIMPNMes,Ad,Me)– with unaltered tripodal ligand. ## with N6 (= inserted nitrogen) instead of C3 (= CCarb.) as third atom to define the plane.
IR Experiments
KBr pellets of the azide complex 7 were irradiated and the changes followed over time by
IR spectroscopy. The majority of the IR features of the cobalt(II) complex is conserved,
with the following exceptions: The azide stretch diminishes, and a new, broad band grows
with its absorption maximum at 1605 cm-1, as well as one at 3337 cm–1 (Figure 48). These
wavenumbers are assignable to an imine bond (C=N) and an N–H stretching frequency,
respectively, so the spectra surely show the gradual conversion of 7 to 11*.
114
3000 2000 1000
0 h
1 h
3 h
5 h
18 h
Tra
nsm
itta
nce
/ %
Wavenumber / cm-1
Figure 48. Photolysis of 7 in a KBr pellet, followed by IR spectroscopy.15
A 15N labeled batch of 7 was prepared with Na15N3, wherein one terminal nitrogen in each
azide molecule is of the heavier isotope. Consequently, in statistically 50 % of the azide
complex, the 15N atom is coordinated directly to the cobalt, which, upon release of
dinitrogen, would result in a 50 % labeled “cobalt-nitrido” or follow-up product. In the IR
spectrum of the labeled compound 7, the azide stretches are shifted slightly towards smaller
wavenumbers (in KBr: 2070, 2035, and 1989 cm–1 vs. 2081, 2044 and 1999 cm–1,
respectively).16 When a KBr pellet of 15N labeled 7 is irradiated, the newly arising bands
are shifted almost negligibly, to 1603 cm–1 (transmission minimum) and 3332 cm–1 (see
spectra p. 209).
This is in line with the assignments of the bands to the cobalt-imino species 11*: The
reduced mass calculation for a harmonic oscillator would predict a higher shift for a
terminal Co-nitride (86 cm–1), for example, but the cobalt-bound imido stretch would shift
15 No smoothing or baseline correction was performed during spectra processing to allow an undistorted
evaluation of the bands’ relative transmittances over time. 16 Instead of two distinctive sets of bands, for the two isotopomers in the batch of 7, all three bands are shifted
entirely since the symmetric and asymmetric stretches involve all three azide nitrogen atoms, not just one
terminal atom of the azide.
115
less since the N is bound between the cobalt and carbene, and therefore the reduced mass
effect is much smaller.
EPR Measurements
While the isolation and characterization of a secondary product following the photolysis
provides strong evidence for the generation of an intermediate cobalt nitride, spectroscopic
confirmation of its generation is desirable to complement the evidence. The direct
photolysis product of a d7 Co(II) azide is expected to be a Co(IV) nitride, a d5 electron
system with presumably an S = ½ ground state. On that premise, EPR spectroscopy can be
a powerful tool to characterize this compound. In addition, 59Co, the only natural isotope,
has a nuclear spin of 7/2, lending a distinctive hyperfine coupling pattern of 8 lines to most
cobalt EPR spectra. Finally, if superhyperfine coupling to one N nucleus could be
observed, this would help to finally identify the species as the nitride.
The following X-band EPR measurements were, if not otherwise mentioned, carried out in
a helium-cooled EPR cavity at temperatures between 8 and 15 K with a LOT 150 W
Xe-OF arc lamp.
The EPR spectrum of the azide complex 7 is of nearly axial symmetry, with effective
g-values around 4.6, 4, and 1.8 – typical of a d7, S = 3/2 system, although no cobalt-
coupling was observed. The sample was irradiated in situ, that is, within the EPR cavity
and under constant cooling by liquid helium. Over time a pronounced, fairly symmetrical
signal with distinct hyperfine coupling evolved around g = 2, indicating a new species with
an S = ½ spin state (Figure 49). This is accompanied by a color change of the irradiated
part of the sample from blue to green (Figure 50).
116
Figure 49. EPR spectra of a sample of 7 (ca. 1.5 mM in frozen methyl-THF) before (black line) and
after (blue line) irradiation with a Xe arc lamp at 10 K. On the right hand side, the progression over
time (in minutes) is shown.
Figure 50. EPR sample of 7 in methyl-THF (ca. 1.5 mM), still frozen, after photolysis at 10 K
within the EPR cavity. The photolyzed part in the middle has turned green, the part that is still blue
was not within the cavity’s quartz glass window and consequently shielded from the lamps’
radiation.
117
The experiment could be reproduced on several different batches of 7 (Figure 51). As soon
as a sample is de-frosted, even for a few seconds, the new signal disappears, leaving behind
only the rhombic signal of (one or more) S = 3/2 species.
Figure 51. EPR spectra of photolyzed samples of 7 (ca. 1.5 mM in methyl-THF) from different
batches of regular 7 (black and blue lines) and one batch of 15N labeled azide (orange line).
Furthermore, it was attempted to reduce the experimental effort and expenditures by
running the photolysis with nitrogen cooling only, keeping the sample at temperatures
around 80 to 88 K, either within the EPR cavity or by photolyzing outside thereof in a
nitrogen-cooled Quartz Dewar (see photo in Figure 107 p. 210). In either case, the spectra
obtained differed significantly from the original one. Not only can the cobalt(II) species not
be observed at 80 K and not only is the signal-to-noise ratio smaller, both of which was to
be expected, but photolysis at 80 K apparently results in more than one species with
g-values around 2 (Figure 52).
118
Figure 52. Samples of 7 (ca. 1.5 mM) in frozen methyl-THF irradiated and measured at 10 K (blue
line) and at 80 K (red and brown lines).
Returning to the 10 K spectra, the main, enveloping eight-lines coupling pattern clearly
demonstrates that the unpaired electron spin must be coupling with one cobalt nucleus. The
smaller features in the center of the spectrum may arise from different sources: A) nitrogen
coupling, which would pinpoint the generation of a Co–N species, B) anisotropic cobalt-
coupling with large and small coupling constants, or C) an underlying second S = ½
species. An aspect in disfavor of option C) is the accurate reproducibility of the experiment
demonstrated in Figure 51: the different species would have to be generated in precisely the
same ratio every time, for different samples and irradiation times.
Simulation of the S = ½ spectrum has proven intricate. One of the best fits obtained so far
(Figure 53) uses cobalt coupling only, speaking for option B) (no N-coupling observable).
However, even in this fairly good fit, not all of the finer features of the spectrum are
reproduced exactly.
119
Figure 53. EPR spectrum (black) and simulation (red) with parameters.
To further examine whether the smaller hyperfine coupling patterns arises from nitrogen
coupling, photolysis experiments were carried out with 15N labeled azide complex.
However, as the labeled azide bears only one 15N isotope at one end of the N3– ligand, and
coordination of the cobalt to the azide would be random to either end, only 50 % of a
generated Co-nitride would be 15N labeled. Furthermore, the nuclear spins of the two
nitrogen isotopes (1 and ½) would cause 3 + 2 lines that would densely overlay each other.
So, all in all, only small changes in the spectrum are expected to begin with.
The spectrum obtained with the 15N labeled sample remained nearly identical to the ones
from non-labeled batches (see Figure 51). Therefore, option B) (only cobalt alone
responsible for the spectral features) seems more likely to be true than option A) (nitrogen
coupling).
Computational Study
To better understand the spectroscopic results and possibly improve the simulated fit of the
EPR spectrum, DFT calculations were performed by Dr. Marat Khusniyarov. The crystal
structure of 7 served as a basis for the calculation, for which an N2 unit was removed from
120
the azide and an S = ½ ground state was assumed. Table 13 summarizes the calculated
structural parameters of the putative nitride complex “[(BIMPNMes,Ad,Me)CoIV(N)]”,
Table 14 lists the calculated EPR parameters and those from the fit in Figure 53. The spin
density map (Figure 54) obtained from these calculations reveals that a large portion of the
unpaired spin is localized on the nitrogen atom. This puts into question the assignment of
the species as a “Co(IV) nitride”, vs. a “Co(III) nitridyl radical”, in analogy to Schneider’s
“iridium(III) nitridyl” complex.[229]
Table 13. Principal bond distances [Å] and an angle [º] of the optimized structure of
“[(BIMPNMes,Ad,Me)CoIV(N)]” obtained from the spin-unrestricted BP-DFT calculations.
Bond / Angle “[(BIMPN
Mes,Ad,Me)Co
IV(N)]”
S = 1/2
Co···Nanchor 3.393
Co–N 1.586
Co–CCarb. 1.913
1.906
Co–O 1.996
N–Co–Nanchor 170.2
121
Table 14. Calculated EPR parameters for “[(BIMPNMes,Ad,Me)CoIV(N)]” (S = 1/2)# and, for
comparison, those of the simulated spectrum in Figure 53; ACo: hyperfine coupling with a 57Co
nucleus; AN: hyperfine coupling with a 14N nitride nucleus. Hyperfine coupling constants are given
in units of MHz.
“[(BIMPNMes,Ad,Me
)CoIV
(N)]”
S = 1/2
simulation parameters
(Figure 53)
g1 1.911 1.992
g2 1.990 2.085
g3 2.008 2.038
giso 1.970
ACo1 –109 80
ACo2 +160 80
ACo3 +545 272
ACo
iso +199
AN1 –18 -
AN2 +28 -
AN3 +48 -
AN
iso +19
# The parameters have been obtained from the spin-unrestricted ZORA-B3LYP-DFT calculations.
Figure 54. Spin density map for “[(BIMPNMes,Ad,Me)CoIV(N)]” (S = 1/2) obtained from the spin-
unrestricted B3LYP-DFT calculations; side view (left) and top view (right) of the complex along
the N–Co–Nanchor axis.
122
As can be easily recognized, the calculated coupling parameters are too large to fit the
experimental spectrum: The cobalt coupling of 545 MHz alone would render the simulated
spectrum too broad. Nevertheless, the result that the nitride coupling is one order of
magnitude smaller than the cobalt one would be in agreement with the assumption that the
nitride coupling is not resolved in the experimental spectrum.
ENDOR Spectroscopy
Since it appears that the nitrogen coupling of the putative CoIV nitrido (or CoIII nitridyl)
species cannot be observed in the EPR experiment, it was tried to determine the nitrogen
coupling constants AN by ENDOR spectroscopy (electron-nuclear double resonance). This
analytical method combines the sensitivity of EPR with the high resolution of NMR
spectroscopy.
A sample of 7 was irradiated in frozen methyl-THF matrix with a Xe flash lamp. During
irradiation, the sample temperature was kept below 10 K. It was then cooled to 5 K and the
spectrum in Figure 55 was taken. The spectrum shows coupling of the unpaired electron to
hydrogen (which may be ligand or solvent hydrogens) and the smallest Co coupling. In
between these two, four lines assignable to N coupling emerge. They are not yet distinctive
enough to confidently evaluate them, but this experiment lays a sound basis for future
experiments.
Still more time needs to be invested to optimize the conditions of photolysis for the
ENDOR measurement. The difficulty is also to find a suitable ENDOR instrument with
both a window in the cavity for light and the potential for liquid helium cooling. Probably,
light source and/or wavelength need to be optimized as well. When the sample was further
irradiated with the Xe flash lamp after the ENDOR spectrum had been taken, the
concentration of the observed “nitrido species” seemed to diminish again.
123
Figure 55. ENDOR spectrum taken at 5 K of a sample of 7 that was irradiated by a xenon flash
lamp.
2.7.3 Photolysis of [(TIMENMes)Co(N3)]+
In the course of his work on TIMENR cobalt imido complexes,[70, 190] Xile Hu treated
[(TIMENXyl)CoI]Cl with several organic azides. When the Co(I) complex was reacted with
TMS-azide in the hope of creating a TMS-imido complex, the one-electron oxidized azido
complex [(TIMENXyl)Co(N3)]Cl was obtained instead.[70] While this complex was
characterized with 1H NMR and IR spectroscopy, elemental and X-ray single crystal
analysis (in crystals of [(TIMENXyl)Co(N3)]BPh4 · MeCN, see Table 7. p. 93), no mention
is made in any of Hu’s works about photolysis experiments on the azide complex, nor any
other means or ways of creating a cobalt nitride.[190] Therefore, after the (BIMPNR,R’,R”)–
azide complexes showed promising potential in this area, photolysis of the TIMENR cobalt
azide complexes and their possible transformation to nitrides was studied.
Synthesis and Characterization of [(TIMENMes
)Co(N3)]+
While Hu’s synthesis of [(TIMENXyl)Co(N3)]Cl involved the sensitive Co(I) complex and
TMS-N3, a more convenient pathway was found: As for the (BIMPNR,R’,R”)– congener, the
azide can synthesized from the well-known Co(II) chlorido complex. And while the
H
Co
124
TIMENR Co(II) chlorido complexes (R = Xylyl, Mesityl) have hitherto been synthesized
by oxidizing the corresponding Co(I) complex, e.g. by benzyl chloride or CH2Cl2,[69, 225] it
has now been found that it can be obtained directly by reacting the free TIMENR ligand
with CoCl2. Also, a direct 1-pot synthesis to the azido complex starting directly with the
+, nine such signals are observed, with two signals of (3+3)
protons intensity probably broadened into the baseline. 18 One signal of 18 H intensity is very broad and probably stems from the merging signals of all ortho-
methyl-groups of the mesityl substituents.
126
Crystal Structure of [(TIMENMes
)Co(N3)]OTs (12[OTs])
Suitable crystals of 12[OTs] for X-ray single crystal analysis were recovered as blue blocks
by cooling an acetonitrile solution to -35 °C. The molecular structure is shown in
Figure 57. The azide complex cations’ structure is very similar to that described by Hu,[70]
with one small but notable difference: The azide in [(TIMENMes)Co(N3)]+ (12
+) is
coordinated more linearly ((Co–Nα–Nβ) = 175.4(2)° in 12+ vs. 166.3(2)° in
[(TIMENXyl)Co(N3)]+), accompanied by a longer Nα–Nβ bond and a shorter Nβ-Nγ bond
(each by 0.1 Å), which again demonstrates the connection between the (linear)
coordination mode of the azide and its activation towards dinitrogen loss. This difference is
naturally more pronounced when compared to the azide complex 7, in which the Co–Nα
difference is also longer by about 0.1 Å.
Figure 57. Molecular structure of the complex [(TIMENMes)Co(N3)]OTs (12[OTs]) in crystals of
[(TIMENMes)Co(N3)]OTs · 3 MeCN (50 % probability ellipsoids). Co-crystallized solvents and
hydrogen atoms are omitted for clarity. Selected bond distances and angles are summarized in
Table 15.
127
Table 15. Selected bond distances [Å], bond angles [°], and doop [Å] with e.s.d.’s in parentheses for
The molecular structures of 11* showed that the complex cation was paired with a chloride
anion, the source of which has not been determined yet. Hence, deliberate supply of
sufficient chloride may also increase the compounds’ yield. For this, however, a chloride
source that is well-soluble in THF would be advantageous. Common chloride salts such as
alkali salts or PNPCl are therefore not purposeful; possibly, an organic ammonium chloride
such as n-Bu4NCl can be used. Alternatively, another anion may contribute to product
isolation, provided it does not absorb too strongly in the crucial area of the electromagnetic
spectrum and it does not cause salt metathesis in 7, i.e. does not eliminate the azide from
the complex. Also, a higher quality crystal for single crystal analysis may be obtained.
It may also be worthwhile to examine the tBu3-phenol’s contribution to the reaction. While
it has been assumed that the phenol provides the hydrogen atom to the imino group, this is
132
still open to discussion. Although by 1H NMR, the phenol does greatly influence the
reactions’ outcome, the phenol may stabilize an intermediate in other ways. Other possible
H-atom sources include the solvent or the ligand. Deuteration experiments may lead to new
insights. Chirik and coworkers proved through deuteration experiments that, in the imine
complex which they obtained upon photolysis of their square planar cobalt azide, the
nitrogen’s proton stems from the ligand.[230]
Furthermore, the essays with other trapping reagents, particularly tert-butyl-isocyanide
(tBuNC), hint that other trapping products may be at hand. The isolobal relationship
between the isocyanide and carbon monoxide suggests that photolysis under a carbon
monoxide (instead of dinitrogen) atmosphere might lead to yet another isolable follow-up
product.
The behavior of other (BIMPNR,R’,R’’)– cobalt azide complexes, 7Xyl,tBu,tBu or 7
Mes,tBu,tBu,
may also be studied. On the one hand, the experience that crystals are far more difficult to
obtain with the tBu-substituted ligand derivatives is somewhat deterring. On the other
hand, these azide complexes may be soluble in aromatic solvents, like their chloride and
PF6 counterparts, and therefore be photolyzed in benzene solution. This could shed a light
on the question whether the solvent influences the reaction – e.g. through THF radicals.
Last but not least, in-depths computational studies may provide crucial additional insights
into the reaction pathways. Comparative computational studies with the TIMENR analogue
12+ may also elucidate whether this compound undergoes the same process, i.e. also forms
an imino insertion product. Comparison of reaction barriers may answer the question
whether the paramagnetic photolysis product of 12+ observed by 1H NMR and EPR can
indeed be a species analogous to 11*.
It is highly doubtable that a cobalt nitrido species can ever be isolated itself, but
spectroscopic evidence of its intermediacy, complemented with computational studies, and
supplemented with isolation and characterization of a secondary product, will provide all
the necessary evidence to prove its fleeting existence. The work is being carried on by Eva
Zolnhofer.
133
3 Summary in English and German
3.1 Summary
Tripodal ligands provide a powerful platform for small molecule activation. The ligand
field splitting resulting from their trigonal coordination environment is suitable for the
stabilization of highly unusual metal-ligand multiple bonds, even for relatively electron
rich mid to late transition metals. Such species can serve as model complexes for
intermediates that have been either postulated or spectroscopically observed in biocatalytic
reactions and industrial catalytic processes, and therefore aid in the elucidation of reaction
mechanisms and in our understanding of chemical transformations in nature. This, in turn,
may aid the development of new catalysts. Furthermore, said species may themselves be
used in atom- and group-transfer chemistry and catalytic transformations, e.g., imido
complexes may be used for aziridination reactions.
The N-anchored, tripodal ligands TIMENR (tris[2-(3-R-imidazol-2-ylidene)ethyl]amine,
R = alkyl, aryl) and ((R,R’ArO)3N)3– (trianion of tris[(3,5-R,R’-2-hydroxyphenyl)methyl]-
amine, R, R’ = alkyl) have been employed in the Meyer group’s laboratories for transition
metal and small molecule activation chemistry. In this work, synthetic routes were
developed towards the mixed bis(carbene)mono(phenolate) ligand (BIMPNR,R’,R’’)–
(anion of bis[2-(3-R-imidazol-2-ylidene)ethyl-(3,5-R’,R”-2-hydroxyphenyl)methyl]amine).
Together with its counterpart, the mono(carbene)bis(phenolate) ligand (MIMPNR,R’,R’’)2–
(dianion of mono[2-(3-R-imidazol-2-ylidene)ethyl]-bis[(3,5-R’,R”-2-hydroxyphenyl)-
methyl]amine), this hybrid ligand bridges the gap between TIMENR and ((R,R’ArO)3N)3–,
creating a complete an N-anchored ligand series with donor functionalities ranging from
tris(carbene) to tris(phenolate) (Chart 3).
NN
N
3
R
TIMENR
N
O
3
((R,R'ArO)3N)3-(BIMPNR,R',R'')-
NN
N
2
RO
(MIMPNR,R',R'')2-
NN
NR
O
2
R'
R'' R''
R' R
R'
Chart 3. Series of tripodal N-anchored ligands from tris(carbene) (left) to tris(phenolate) (right).
134
One synthetic route towards (BIMPNR,R’,R’’)–, in which the phenol is attached first to the
nitrogen anchor, involves a tosylation reaction in order to create a highly reactive leaving
group for the following nucleophilic substitution with the imidazole (Scheme 49). This
route may be employed particularly for (BIMPNR,R’,R’’)– derivatives in which the imidazole
building block is more precious than the phenolate.
Mannich reaction
OH
R''
R'
NN
NR
2
OH
R'
R''
NN
NR
2
OK
R'
R''
HNOH
2 NHO
2
OH
R'
R''
ClOTs
tosylation
SN2
2N
N
R
KOtBu
deprot.
OTs
NTsO
2
OH
R'
R''
K(BIMPNR,R',R'')(H3BIMPNR,R',R'')(OTs)2 Scheme 49. Synthetic route for K(BIMPNR,R’,R’’)– in which the phenol is attached first to the
N-anchor.
Another route with better overall yields and easier workup was developed in collaboration
with Johannes Hohenberger, who researches the (MIMPNR,R’,R”)2– ligand and its
coordination chemistry. This route is used for more easily prepared imidazoles such as
xylyl- and mesityl-imidazoles. Scheme 50 summarizes this synthesis for both new, hybrid
ligand types (BIMPNR,R’,R”)– and (MIMPNR,R’,R”)2–: In a first step, the carbene units are
attached to the nitrogen anchor by SN2 reaction of a halido-ethylamine and the substituted
imidazol. The phenol is chloromethylated via a Blanc reaction, after which the amine
nitrogen nucleophilically attacks the benzylic position to form the ligand precursors
(H3BIMPNR,R’,R”)2+ and (H3MIMPNR,R’,R”)+, respectively. These ligand precursors are
deprotonated with KOtBu in THF to give the potassium salts of the free ligands in near
quantitative yield.
135
OH
R'
R''
H3-nNX
n
x HXn
N
N
R
H3-nNN
x HX
NR
n
X
(X = Cl, Br)
OH
R'
R''
Cl
NN
NR
n
OH
R'
R''3-n
(toluene)
HCl conc.
n = 2: (H3BIMPNR,R',R'')2+
n = 1: (H3MIMPNR,R',R'')+
NN
NR
n
OK
R'
R''3-n
n = 2: K(BIMPNR,R',R'')
n = 1: K2(MIMPNR,R',R'')
KOtBu
CH2O
Scheme 50. Synthetic route for K(BIMPNR,R’,R’’) and K2(MIMPNR,R’,R’’).
Both the protonated ligand precursor and the ligand’s potassium salt have been
characterized thoroughly, not only by elemental analysis, 1H, 13C, and 2D NMR as well as
IR spectroscopy, but also by X-ray single crystal analysis.
The complete ligand series offers great tunability of the electronic and steric environment
around the metal center, allowing adjustment of their complexes’ reactivity. Additionally,
the modular synthesis of the mixed ligands allows combination of different substituents on
the NHC and phenolate moieties.
The coordination of the novel, chelating tripodal ligands to Mn, Fe and Co was explored
and a range of divalent complexes of Mn, Fe, and Co was synthesized and characterized by 1H NMR, IR and UV/Vis spectroscopy as well as single crystal X-ray diffraction. Variable
temperature SQUID magnetization measurements in the range from 2 to 300 K confirmed
high spin ground states for the divalent complexes. 57Fe Mößbauer spectroscopy of Fe(II)
complexes 1, 1[Br], 1[BPh4], 1Mes,tBu,tBu and 2 in comparison with data of corresponding
TIMENR complexes revealed an increase of the isomer shifts δ of about 0.1 mm · s–1
caused by the substitution of one (σ-donating/ π-backbonding) NHC with one (σ- and
π-donating) phenolate.
136
Figure 61. Molecular structures of divalent complexes of (BIMPNR,R’,R’’); top row: chlorido
complexes of the formula [(BIMPNMes,Ad,Me)MIICl], bottom row: azide complexes of iron and cobalt
and [(BIMPNXyl,tBu,tBu)Co]PF6 (6Xyl,tBu,tBu[PF6]). The iron complex 1 is coordinated by an
acetonitrile molecule (1(MeCN)) after crystals were grown from an acetonitrile solution.
The complexes’ crystal structures (Figure 61) reveal the different steric demand of the new
ligand. Particularly, the molecular structure of 3 – in which a pyridine molecule is situated
next to the Mn–Cl bond – and those of azide complexes 2 and 7, in which the azide is
coordinated in its preferred bent coordination mode, demonstrate the mixed ligand’s
flexibility, and, in contrast to the corresponding TIMENR ligands, their potential to allow
side access to the reactive center for e.g. organic substrates.
Cyclovoltammograms on the divalent complexes feature an unusually large wave
separation for the redox-wave assigned to the metal centered MII/MIII oxidation, which is
explainable with a rearrangement of the coordination sphere upon electron transfer. The
isolated and spectroscopically characterized diamagnetic complex [(BIMPNMes,Ad,Me)Zn]X
137
(5 and 5[OTs], X = Cl, OTs) may serve as a reference complex with non-redox-active
metal center for further electrochemical studies.
In place of Hu’s way of synthesizing TIMENR Co(I) complexes treating the precursor
CoCl(PPh3)3 with the free ligand, it was found that the reduction of the cobalt(II) complex
was a much more favorable route for the (BIMPNR,R’,R’’)– ligand system. Reaction of the
reddish-brown cobalt(I) complex [(BIMPNMes,Ad,Me)CoI] (8) (Scheme 51) with organic
azides led to different outcomes depending on the size of the azide’s substituent: With trityl
azide, 8 reacted to 7, demonstrating that the trityl substituent is too large for the reactive
cavity, and rather releases its N3– group instead. Treatment of 8 with mesityl azide (MesN3)
led to the green compound [(BIMPNMes,Ad,Me*N3Mes)CoII] (10*), in which the whole
MesN3 unit is bound to one of the ligand’s NHC arms at the former carbene carbon atom.
The NHC ring is flipped around and is now coordinated as an alkenyl, sometimes referred
to as “abnormal carbene”. With phenyl azide (PhN3), effervescence is observed, and a
green compound is obtained. In the violet crystals grown from solutions of this green
compound, the dinuclear complex [bis-µ-OH{Co(BIMPNMes,Ad,Me*NPh)}2] (9*) was found.
Mechanistic considerations led to the conclusion that solely those aryl azides without
ortho-substituents are capable of binding to the cobalt center in 8 through the aryl-bound
nitrogen atom, i.e., in the “bent” coordination mode necessary for release of dinitrogen.
138
Scheme 51. Reaction of cobalt(I) complex 8 with aryl azides: Mesityl azide leads to abnormal
carbene 10*, phenyl azide to a putative, intermediate imido complex and ultimately to the dinuclear
bis-µ-hydroxo complex 9*.
In the search for an isolable cobalt nitride complex, the photolysis behavior of the blue
azide complexes [(BIMPNMes,Ad,Me)Co(N3)] (7) and [(TIMENMes)Co(N3)]X (12[X], X =
OTs, PF6) was studied. When 7 is photolyzed in frozen solution at very low temperatures
(10 K), an X-band EPR signal emerges around g = 2 with distinctive hyperfine coupling
pattern, indicative of an S = ½ cobalt species (Figure 62). Irradiation at liquid nitrogen
temperatures (80 – 86 K) leads to less well-defined spectra which apparently originate from
at least two species, and the signal vanishes entirely if the solution is defrosted even for a
second. All of this proves the extremely low stability and/or high reactivity of the observed
species. Experiments with trapping reagents were conducted in an effort to isolate a
secondary product following the photolysis of 7. Photolysis in the presence of 2,4,6-tris-
tert-butylphenol led to the green imino complex [(BIMPNMes,Ad,Me*NH)CoII]Cl (11*)
139
(Figure 47), which corroborates the assignment of the S = ½ EPR spectrum to a cobalt-
nitrido species. The proposed reaction mechanism towards 11* and intermediacy of the
nitride species is supported by IR, 1H NMR and preliminary ENDOR data.
Figure 62. X-band EPR signal obtained after photolysis of 7 at 10 K in frozen methyl-THF
solution, left: spectrum (black line) and simulated fit (red line) with parameters, right: growth of the
signal with irradiation time (given in minutes).
Figure 63. Molecular structure of the insertion product [(BIMPNMes,Ad,Me*NH)Co]Cl (11*).
140
Hu’s synthesis of [(TIMENXyl)Co(N3)]Cl involved the highly reactive TIMENXyl cobalt(I)
complex and TMS-N3. For the TIMENMes azide complex 12[X], two more convenient
routes were developed: Firstly, salt metathesis from the corresponding chloride complex,
and secondly, direct synthesis by reaction of the free ligand with CoCl2, an excess of NaN3,
and 1 equivalent of NaX (X = OTs, PF6). Photolysis of 12[X] in acetonitrile solution
apparently led to a well-defined paramagnetic species (see Figure 64). However, EPR
experiments at low temperatures have not permitted the observation of an intermediate
S = ½ species. The elucidation of this photolysis product’s identity is intriguing both in
itself and with regard to formation of 11*, i.e. to see in how far the photolytic behavior of
12[X] is analogous to or different from the photolytic behavior of 7.
Figure 64. 1H NMR spectrum of an irradiated sample of 12[OTs] in acetonitrile-d3 after full
conversion of the azide complex.
141
3.2 Zusammenfassung
Tripodale Liganden bieten eine leistungsstarke Plattform zur Aktivierung kleiner Moleküle.
Ihre trigonale Koordinationssphäre führt zu einer Ligandenfeldaufspaltung, die zur
Stabilisierung ungewöhnlicher Metall-Ligand-Mehrfachbindungen selbst bei relativ
elektronenreichen, mittleren bis späten Übergangsmetallen genutzt werden kann. Solche
Spezies dienen als Modellkomplexe von Intermediaten, welche in biokatalytischen
Reaktionen ebenso wie in industriellen Anwendungen entweder postuliert oder
spektroskopisch beobachtet wurden. Somit können sie zur Aufklärung von Reaktions-
mechanismen beitragen und unser Verständnis über chemische Umwandlungen in der
Natur erweitern. Dies kann wiederum die Entwicklung neuer Katalysatoren voranbringen.
Ferner können besagte Spezies selbst in Atom- und Gruppen-Transferreaktionen eingesetzt
werden, beispielsweise werden Imidokomplexe zu Aziridinierungen genutzt.
Die stickstoffgeankerten, tripodalen Liganden TIMENR (Tris[2-(3-R-imidazol-2-yliden)-
ethyl]amin, R = Alkyl, Aryl) und ((R,R’ArO)3N)3– (Trianion von Tris[(3,5-R,R’-2-
hydroxyphenyl)methyl]amin, R, R’ = Alkyl) werden in den Laboren des Arbeitskreises
Prof. Meyer für Übergangsmetallchemie und zur Aktivierung kleiner Moleküle eingesetzt.
In dieser Arbeit wurden Synthesewege hin zum gemischten Bis(carben)mono(phenolat)-
Liganden (BIMPNR,R’,R’’)– (Anion von Bis[2-(3-R-imidazol-2-yliden)ethyl]-[(3,5-R’,R”-2-
hydroxyphenyl)methyl]amin) entwickelt. Zusammen mit seinem Gegenstück, dem
Mono(carben)bis(phenolat)-Liganden (MIMPNR,R’,R’’)2– (Dianion von Mono[2-(3-R-
imidazol-2-yliden)ethyl]-bis[(3,5-R’,R”-2-hydroxyphenyl)methyl]amin), schließt dieser
Hybridligand die Lücke zwischen TIMENR und ((R,R’ArO)3N)3– und schafft somit eine
vollständige Ligandenserie mit Donor-Gruppen von Tris(carben) bis Tris(phenolat)
(Schema 1).
NN
N
3
R
TIMENR
N
O
3
((R,R'ArO)3N)3-(BIMPNR,R',R'')-
NN
N
2
RO
(MIMPNR,R',R'')2-
NN
NR
O
2
R'
R'' R''
R' R
R'
Schema 1. Serie tripodaler, stickstoffgeankerter Liganden von Tris(carben) (links) bis
Tris(phenolat) (rechts).
142
In einer der (BIMPNR,R’,R’’)– Syntheserouten, in welcher das Phenol als erstes an den
Stickstoffanker gekoppelt wird, wird durch einen Tosylierungsschritt um eine hochreaktive
Abgangsgruppe geschaffen für die darauffolgende nukleophile Substitution durch das
Imidazol (Schema 2). Dieser Weg ist besonders für den Einsatz bei (BIMPNR,R’,R’’)-
Derivaten mit wertvollen Imidazolen geeignet.
Mannich- Reaktion
OH
R''
R'
NN
NR
2
OH
R'
R''
NN
NR
2
OK
R'
R''
HNOH
2 NHO
2
OH
R'
R''
ClOTs
Tosylierung
SN2
2N
N
R
KOtBu
Deprot.
OTs
NTsO
2
OH
R'
R''
K(BIMPNR,R',R'')(H3BIMPNR,R',R'')(OTs)2 Schema 2. Syntheseroute für K(BIMPNR,R’,R’’), in welcher das Phenol zuerst am Stickstoffanker
angebracht wird.
Eine weitere Route mit besserer Gesamtausbeute und einfacherer Aufarbeitung wurde in
Zusammenarbeit mit Johannes Hohenberger entwickelt, welcher das (MIMPNR.R’,R’’)2–
System erforscht. Diese Route wird bei leichter herzustellenden Imidazolen wie jenen mit
Xylyl- und Mesityl-Substituenten verwendet. Schema 3 fasst diese Synthese für beide
neuen Ligandentypen, (BIMPNR.R’,R’’)– und (MIMPNR.R’,R’’)2–, zusammen: Im ersten Schritt
werden die späteren Carben-Einheiten am Stickstoffanker fixiert durch SN2-Reaktion
zwischen halogeniertem Ethylamin und Imidazol. In einer Blanc Reaktion wird das Phenol
chlormethyliert, woraufhin der Aminstickstoff die benzylische Position nukleophil
angreifen kann, um zu den Ligandenvorstufen (H3BIMPNR.R’,R’’)2+ und (H3MIMPNR.R’,R’’)+
zu gelangen. Diese werden in THF mit KOtBu deprotoniert, woraufhin man das
Kaliumsalz der freien Liganden in nahezu quantitativer Ausbeute erhält.
Beide protonierten Ligandenvorstufen sowie das Kaliumsalz des Liganden wurden
eingehend charakterisiert, nicht nur mittels Elementaranalyse, 1H, 13C und 2D NMR- sowie
IR-Spektroskopie, sondern auch durch Röntgenstrukturanalyse am Einkristall.
143
OH
R'
R''
H3-nNX
n
x HXn
N
N
R
H3-nNN
x HX
NR
n
X
(X = Cl, Br)
OH
R'
R''
Cl
NN
NR
n
OH
R'
R''3-n
(toluene)
HCl conc.
n = 2: (H3BIMPNR,R',R'')2+
n = 1: (H3MIMPNR,R',R'')+
NN
NR
n
OK
R'
R''3-n
n = 2: K(BIMPNR,R',R'')
n = 1: K2(MIMPNR,R',R'')
KOtBu
CH2O
Schema 3. Syntheseroute für K(BIMPNR,R’,R’’) und K2(MIMPNR,R’,R’’).
Die elektronische und sterische Umgebung am Metallzentrum kann durch die komplette
Ligandenserie einfach eingestellt werden, wodurch die Reaktivität der resultierenden
Komplexe an den jeweiligen Zweck angepasst werden kann. Zusätzlich erlaubt die
bausteinartige Synthese der gemischten Liganden freie Kombination verschiedener
Substituenten an den NHC- und Phenolat-Ringen.
Die Koordination der neuartigen Chelat-Liganden an Mn, Fe und Co wurde untersucht und
eine Reihe von zweiwertigen Komplexen dieser Metalle synthetisiert und mittels 1H NMR-, IR- und UV/Vis-Spektroskopie sowie Einkristall-Röntgenstrukturanalyse
charakterisiert. SQUID Magnetisierungsmessungen im Temperaturbereich von 2 bis 300 K
bestätigten high spin Grundzustände für die zweiwertigen Komplexe. Der Vergleich der 57Fe Mößbauer Spektren der Eisen(II)-Komplexe 1, 1[Br], 1[BPh4], 1
Mes,tBu,tBu und 2 mit
jenen der entsprechenden TIMENR-Komplexe zeigte, dass sich die Isomerieverschiebung δ
um etwa 0.1 mm · s–1 erhöht, hervorgerufen durch den Austausch eines (σ-donierenden, π-
rückbindenen) NHCs durch ein (σ- und π-donierendes) Phenolat.
Abbildung 4. 1H NMR Spektrum einer bestrahlten Probe von 12[Ots] in Acetonitril-d3 nach
vollständigem Umsatz des Azidkomplexes.
149
4 Experimental Part
4.1 Methods, Procedures and Starting Materials
4.1.1 General
All air- and moisture-sensitive experiments were performed under dry nitrogen atmosphere
using standard Schlenk techniques or an MBraun inert-gas glovebox containing an
atmosphere of purified dinitrogen.
Solvents for air- and moisture-sensitive experiments were purified using a two-column
solid-state purification system (Glasscontour System, Irvine, CA) and transferred to the
glovebox without exposure to air. NMR solvents were obtained packaged under argon and
stored over activated molecular sieves and sodium (where appropriate) prior to use.
4.1.2 Starting Materials
Manganese(II) chloride and iron(II) chloride, each anhydrous 99.9 %, cobalt(II) chloride,
anhydrous 97 %, and PNPCl 97 % were purchased from Sigma-Aldrich and used as
received. Sodium azide 97 % and sodium hexafluorophosphate were purchased from
ACROS Organics, as were any organic chemicals for ligand synthesis, and used without
further purification. KOtBu 98+% was obtained from ACROS Organics and purified by
sublimation under reduced pressure before it was transferred into the glovebox. NaOMe
97 % was purchased from Merck and used as received.
Imidazoles[186], 2-adamantyl-para-cresol[187], aryl azides[235] and the cobalt(I) precursor
CoCl(PPh3)3[236] were synthesized according to literature procedures; some other steps in
ligand syntheses were modified from literature procedures which are referenced in the
respective section.
4.1.3 Analytical Methods
1H-NMR spectra were recorded on JEOL 270 and 400 MHz instruments, operating at
respective frequencies of 269.714 and 400.178 MHz with a probe temperature of 23 ºC. 13C NMR spectra were recorded on JEOL 270 and 400 MHz instruments, operating at
respective frequencies of 67.82 MHz and 100.624 MHz with a probe temperature of 23 ºC.
150
Chemical shifts are reported relative to the peak for SiMe4, using 1H (residual) chemical
shifts of the solvent as a secondary standard[237] and are reported in ppm.
Elemental analysis results were obtained from the Analytical Laboratories at the
(s (br), 3 H), 26.40 (s (vbr), 2 H), 21.34 (s, 2 H), 8.26 (s (vbr), 9 H), 1.63
(s, 6 H), -0.48 (s, 6 H), -1.59 (s, 12 H), -2.79 (s, 6 H); no further signals
are observed from 170 to -130 ppm; some broad singlets may be
superimposed and the signals of two protons broadened into the baseline. 1H NMR (399.78 MHz, CDCl3, ppm): δ = 55.65 (s, 2 H), 54.88 (s, 6 H), 33.1 (s
(vbr), 2 H), 28.4 (s (vbr), 3 H), 22.19 (s, 2 H), 5.50 (s (vbr), 9 H), 0.63 (s,
6 H), -1.34 (s, 6 H), -1.98 (s, 12 H), -4.20 (s, 6 H); no further signals are
observed from 170 to -130 ppm; some broad singlets may be
superimposed and the signals of two protons broadened into the baseline. 1H NMR (399.78 MHz, THF-d8, ppm): δ = 91.90 (s, 1 H), 73.29 (s, 1 H), 57.44 (s,
# corresponding angle in complexes of (BIMPNMes,Ad,Me)– with unaltered tripodal ligand. ## with C4 (=CCarb2) instead of C3 (= CCarb.) as third atom to define the plane.
234
Table 32. Selected bond distances [Å] with e.s.d.’s in parentheses for [(BIMPNMes,Ad,Me*NH)Co]Cl
· THF · 0.16 H2O (with the actual compound distribution in the crystal:
denoted as 11* · THF · 0.16 H2O). See Figure 47 for atom labeling.
Angle# Angle 11*
· THF · 0.16 H2O
C3–N6–Co1 130.4(2)
Ccarb.–M–C’carb. N6–Co1–C8 109.9(2)
Ccarb.–M–O C3–Co1–O1
C8–Co1–O1
-
118.5(1)
N6–Co–O1 118.91(9)
NNHC1–Ccarb.–NNHC2 N2–C3–N3
N4–C8–N5
106.5(2)
103.7(2)
doop doop
–0.404 ##
#corresponding angle in complexes of (BIMPNMes,Ad,Me)– with unaltered tripodal ligand. ## with N6 (= inserted nitrogen) instead of C3 (= CCarb.) as third atom to define the plane.
236
Table 34. Selected bond distances [Å] with e.s.d.’s in parentheses for [(TIMENMes)Co(N3)]OTs
· 3 MeCN (12[OTs] · 3 MeCN). See Chart 2 for atom labeling. For swift and easy comparison,
values of [(BIMPNMes,Ad,Me)Co(N3)] (7, see above) and [(TIMENXyl)Co(N3)]BPh4 · MeCN are also
tabulated here.
Bond 7 · 2 THF
· 0.5 C6H6
12[OTs]
· 3 MeCN
[(TIMENXyl)Co(N3)]BPh4
· MeCN
M···Nanchor 2.659 3.234 3.213
M–Laxial 2.039(2) 1.939(2) 1.938(2)
M–Ccarb. 2.081(2)
2.060(2)
2.053(2)
2.058(2)
2.062(2)
2.052(2)
2.049(2)
2.017(2)
M–O 1.931(2) - -
Nα–Nβ 1.188(3) 1.171(3) 1.161(3)
Nβ–Nγ 1.164(3) 1.157(3) 1.169(3)
NNHC1–Ccarb. 1.357(3)
1.354(3)
1.357(3)
1.363(3)
1.358(3)
1.357(3)
1.356(3)
1.358(3)
NNHC2–Ccarb. 1.361(3)
1.360(3)
1.361(3)
1.364(3)
1.366(3)
1.368(3)
1.364(3)
1.361(3)
NNHC1–Ccarb2 1.377(3)
1.378(3)
1.385(3)
1.385(3)
1.387(3)
1.384(3)
1.385(3)
1.382(3)
Ccarb2–Ccarb3 1.338(3)
1.340(3)
1.343(4)
1.342(4)
1.338(4)
1.336(3)
1.339(3)
1.335(3)
NNHC2–Ccarb3 1.385(3)
1.383(3)
1.384(3)
1.391(3)
1.389(3)
1.384(3)
1.382(3)
1.383(3)
237
Table 35. Selected bond angles [°] and doop [Å] with e.s.d.’s in parentheses for
[(TIMENMes)Co(N3)]OTs · 3 MeCN (12[OTs] · 3 MeCN). See Chart 2 for atom labeling. For swift
and easy comparison, values of [(BIMPNMes,Ad,Me)Co(N3)] (7, see above) and
[(TIMENXyl)Co(N3)]BPh4 · MeCN are also tabulated here.
Angle 7 · 2 THF
· 0.5 C6H6
12[OTs]
· 3 MeCN
[(TIMENXyl)Co(N3)]BPh4
· MeCN
Nanchor–M– Laxial 168.7 178.5 174.27
Ccarb.–M–Laxial 102.23(8)
105.10(8)
103.44(9)
104.83(9)
106.02(9)
102.17(8)
101.85(8)
110.48(8)
O–M–Laxial 89.93(7) - -
M–Nα–Nβ 128.2(2) 175.4(2) 166.3(2)
Nα–Nβ–Nγ 177.1(2) 179.5(3) 178.3(2)
Ccarb.–M–C’carb. 106.95(8) 113.62(9)
110.29(8)
117.30(9)
118.86(8)
111.91(8)
110.52(8)
Ccarb.–M–O 134.03(7)
112.31(7)
- -
NNHC1–Ccarb.–
NNHC2
102.8(2)
103.1(2)
103.1(2)
103.2(2)
103.1(2)
103.2(2)
103.6(2)
103.3(2)
doop 0.297(2) 0.524 0.520
238
4.4.4 Crystal Structure of [(BIMPNMes,Ad,Me)Co]PF6 (6[PF6])
Figure 109. Molecular structure of 6[PF6] in crystals of [(BIMPNMes,Ad,Me)Co]PF6 · 3 THF
(50 % probability ellipsoids, hydrogen atoms and solvent molecules omitted for clarity).
239
5 Symbols and Abbreviations
× adduct with
Ad Adamantyl
aq. aqueous
Ar aryl
(BIMPNR,R’,R’’)– anion of bis[2-(3-R-imidazol-2-ylidene)ethyl]-[(3,5-R’,R”-2-hydroxy-
phenyl)methyl]amine
Bn benzyl
br broad
calcd. calculated
δ a) NMR chemical shift
b) Mößbauer chemical isomer shift
d doublet
D zero field splitting parameter
doop
out of plane shift (distance of the metal center from the least-squares
plane defined by the three coordinating atoms of the tripodal ligand;
i.e., the carbene carbon and phenolate oxygen atoms)
∆EQ Mößbauer quadrupole splitting parameter
e.s.d. estimated standard deviation (standard uncertainty)
EA elemental analysis
EPR Electron Paramagnetic Resonance
eq equivalent(s)
Et ethyl
Et2O diethyl ether
FD-MS field desorption mass spectrometry
FWHM full-width at half-maximum
ΓFWHM line width; FWHM = full-width at half-maximum
g EPR g-value
HOMO Highest Occupied Molecular Orbital
240
i. vac. in vacuo (in/ under vacuum)
iPr iso-Propyl
IR infrared
KOtBu potassium tert-butanolate
LUMO Lowest Unoccupied Molecular Orbital
µB Bohr magneton
µeff effective magnetic moment
µmol micromole
m multiplet
M molar
Me methyl
Mes mesityl
(MIMPNR,R’,R’’)2– dianion of mono[2-(3-R-imidazol-2-ylidene)ethyl]-bis[(3,5-R’,R”-2-
hydroxyphenyl)methyl]amine
min. minute(s)
mM millimolar
mmol millimole
MS mass spectrometry
NHC N-heterocyclic carbene
NHE normal hydrogen electrode
NMR Nuclear Magnetic Resonance
obsd. observed
–OTs tosylate = anion of para-toluene-sulfonic acid
PG protecting group
Ph phenyl
py pyridine
quant. quantitative yield
quart. quaternary
RT room temperature
s singlet
241
S total spin angular momentum
sat. saturated
sept septet
SN2 bimolecular nucleophilic substitution
SQUID superconducting quantum interference device
t triplet
tBu tert-butyl
tBu3-phenol 2,3,6-tris-tert-butylphenol
TEMPO tetramethylpiperidinoxyl
tert tertiary
TIMENR tris[2-(3-R-imidazol-2-ylidene)ethyl]amine
TLC thin layer chromatography
TMS trimethylsilyl
Tol tolyl
trityl triphenylmethyl
Ts tosyl = para-toluene-sulfonyl
UV/Vis Ultra-Violet/Visible
ν~ as asymmetric vibration
ν~ wave number
vbr very broad
Xyl xylyl
3,5-Xyl 3,5-xylyl
242
6 Numbered Compounds
Organic compounds
R'
R''
OH
NHO
2
R'
R''
OH
NCl
2
c1-R’,R’’ c2-R’,R’’
R'
R''
OH
NI
2
R'
R''
OH
NTsO
2
c3-R’,R’’ c4-R’,R’’
H2NCl
2
Cl 3 Cl
H3NN
NR
2
p1×HCl p2-R×HCl
HNN
NR
2
2 PF6
OH
R''
R'Cl
p2-R[PF6] p3-R’,R’’
243
Complexes
[(BIMPNMes,Ad,Me)Fe(Cl)] 1
[(BIMPNMes,tBu,tBu)Fe(Cl)] 1Mes,tBu,tBu
[(BIMPNMes,tBu,tBu)Fe(Cl)] 1Xyl,tBu,tBu
[(BIMPNMes,Ad,Me)Fe(Br)] 1[Br]
[(BIMPNMes,Ad,Me)Fe]BPh4 1[BPh4]
[(BIMPNMes,Ad,Me)Fe(N3)] 2
[(BIMPNMes,Ad,Me)Mn(Cl)] 3
[(BIMPNMes,Ad,Me)Mn(N3)] 4
[(BIMPNMes,Ad,Me)Zn(Cl)] 5
[(BIMPNMes,Ad,Me)Zn]OTs 5[OTs]
[(BIMPNMes,Ad,Me)Co]Cl 6
[(BIMPNMes,tBu,tBu)Co]Cl 6Mes,tBu,tBu
[(BIMPNXyl,tBu,tBu)Co]Cl 6Xyl,tBu,tBu
[(BIMPNMes,Ad,Me)Co]PF6 6[PF6]
[(BIMPNXyl,tBu,tBu)Co]PF6 6Xyl,tBu,tBu
[PF6]
[(BIMPNMes,Ad,Me)Co(N3)] 7
[(BIMPNMes,Ad,Me)CoI] 8
[bis-µ-OH{Co(BIMPNMes,Ad,Me*NPh)}2] 9*
[(BIMPNMes,Ad,Me*N3Mes)CoII] 10*
[(BIMPNMes,Ad,Me*NH)CoII]Cl 11*
[(TIMENMes)Co(N3)]OTs 12[OTs]
[(TIMENMes)Co(N3)]PF6 12[PF6]
244
7 References
[1] Hollemann, Wiberg, Lehrbuch der Anorganischen Chemie, 102. Auflage, Walter de
Gruyter, Berlin, 2007, pp. 653 and 898.
[2] European Commission, Joint Research Centre (JRC)/PBL Netherlands
Environmental Assessment Agency. Emission Database for Global Atmospheric
Research (EDGAR), release version 4.2. http://edgar.jrc.ec.europe.eu, 2011.
[3] J. A. Glaser, Clean Technologies and Environmental Policy 2009, 11, 253-257.
[4] E. A. Quadrelli, G. Centi, J.-L. Duplan, S. Perathoner, ChemSusChem 2011, 4,
1194-1215.
[5] C. Das Neves Gomes, O. Jacquet, C. Villiers, P. Thuery, M. Ephritikhine, T. Cantat,
Angew Chem Int Ed Engl 2012, 51, 187-190.
[6] O. Einsle, F. A. Tezcan, S. L. Andrade, B. Schmid, M. Yoshida, J. B. Howard, D.
C. Rees, Science 2002, 297, 1696-1700.
[7] T. Spatzal, M. Aksoyoglu, L. Zhang, S. L. A. Andrade, E. Schleicher, S. Weber, D.
C. Rees, O. Einsle, Science 2011, 334, 940.
[8] J. Postgate, Nitrogen Fixation, 3rd ed., Cambridge University Press, Cambridge
(UK), 1998.
[9] Y. Lee, N. P. Mankad, J. C. Peters, Nat Chem 2010, 2, 558-565.
[10] NITROGEN (FIXED)—AMMONIA, U.S. Geological Survey, Mineral Commodity
Summaries, 2012.
[11] G. Ertl, Angew. Chem. 1990, 102, 1258-1266.
[12] G. Ertl, The Chemical Record 2001, 1, 33-45.
[13] Y. K. J.-R. N. Umena, Nature 2011, 473, 55-60.
[14] T. A. Betley, Y. Surendranath, M. V. Childress, G. E. Alliger, R. Fu, C. C.
Cummins, D. G. Nocera, Philos Trans R Soc Lond B Biol Sci 2008, 363, 1293-
1303; discussion 1303.
[15] T. Privalov, L. Sun, B. Åkermark, J. Liu, Y. Gao, M. Wang, Inorg. Chem. 2007, 46,
7075-7086.
245
[16] Y. Gao, T. Åkermark, J. Liu, L. Sun, B. Åkermark, J. Am. Chem. Soc. 2009, 131,
8726-8727.
[17] N. Cox, D. A. Pantazis, F. Neese, W. Lubitz, Acc Chem Res 2013.
[18] P. E. Siegbahn, Biochim Biophys Acta 2012.
[19] J. Hohenberger, K. Ray, K. Meyer, Nat Commun 2012, 3, 720.
[20] E. R. King, E. T. Hennessy, T. A. Betley, J Am Chem Soc 2011, 133, 4917-4923.
[21] G. Parkin, Chem. Commun. 2000, 1971-1985.
[22] N. Burzlaff, Tripodal N,N,O-ligands for metalloenzyme models and
organometallics in Adv. Inorg. Chem., Vol. Volume 60 (Ed.: R. van Eldik),
Academic Press, 2008, pp. 101-165.
[23] S. E. Gibson, M. P. Castaldi, Angewandte Chemie International Edition 2006, 45,
4718-4720.
[24] S. Trofimenko, Prog. Inorg. Chem. 1986, 34, 115-210.
[25] S. Trofimenko, Chem. Rev. 1993, 93, 943-980.
[26] S. Trofimenko, J. Am. Chem. Soc. 1966, 88, 1842-1844.
[27] S. Trofimenko, Polyhedron 2004, 23, 197-203.
[28] D. L. Reger, C. A. Little, A. L. Rheingold, M. Lam, L. M. Liable-Sands, B.
Rhagitan, T. Concolino, A. Mohan, G. J. Long, V. Briois, F. Grandjean, Inorg.
Chem. 2001, 40, 1508-1520.
[29] M. A. Schwartz, J. Cell Sci. 2008, 121, 1771.
[30] H. Pfeiffer, A. Rojas, J. Niesel, U. Schatzschneider, Dalton Trans. 2009, 0, 4292-
4298.
[31] S. Trofimenko, Scorpionates: The Coordination Chemistry of Polypyrazolylborate
Ligands, Imperial College Press, London, 1999.
[32] R. R. Schrock, Acc. Chem. Res. 1997, 30, 9-16.
[33] D. V. Yandulov, R. R. Schrock, Science 2003, 301, 76-78.
[34] R. R. Schrock, Acc. Chem. Res. 2005, 38, 955-962.
[35] R. R. Schrock, Catalytic Reduction of Dinitrogen to Ammonia by Molybdenum in