Insights in the mechanism of selective olefin oligomerisation catalysis using stopped-flow freeze-quench techniques: A Mo K-edge QEXAFS study

Article abstract of DOI:10.1016/j.jcat.2011.10.015

The activation of [MoX₃(L)] (with X = Cl, Br; L = tridentate ligands with S₃ and SNS donor sets) by AlMe₃, analogous to the industrially important [CrCl₃(L)] catalysts for selective oligomerisation of alkenes, h

Full text of DOI:10.1016/j.jcat.2011.10.015

Journal of Catalysis 284 (2011) 247–258
Contents lists available at SciVerse ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Insights in the mechanism of selective olefin oligomerisation catalysis using
stopped-flow freeze-quench techniques: A Mo K-edge QEXAFS study
Stuart A. Bartlett a, Peter P. Wells a, Maarten Nachtegaal b, Andrew J. Dent c, Giannantonio Cibin c,
Gillian Reid a, John Evans a, Moniek Tromp d,
⇑
a School of Chemistry, University of Southampton, Highfield, Southampton SO17 1BJ, United Kingdom
b Paul Scherrer Institut, 5232 Villigen, Switzerland
c Diamond Light Source Ltd., Didcot OX11 0DE, United Kingdom
d Strukturanalytik in der Katalyse, Chemie, Catalysis Research Center, Technische Universität München, Lichtenbergstrasse 4, 85748 Garching, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
Available online 13 November 2011
The activation of [MoX3(L)] (with X = Cl, Br; L = tridentate ligands with S3 and SNS donor sets) by AlMe3,
analogous to the industrially important [CrCl3(L)] catalysts for selective oligomerisation of alkenes, has
been investigated by Mo K-edge X-ray absorption (XAS) and UV–visible spectroscopies. Time-resolved
stopped-flow XAS, in combination with a newly developed anaerobic freeze-quench approach, have
established the complete alkylation of the Mo centres and a slower, stepwise sequence for [MoBr3(L)].
No evidence for directly bonded or bridged Mo–Mo dimers was observed at the high Mo:AlMe3 ratios
used in this study. Decomposition of the complexes is in competition with the activation and resulted
in precipitation of particulate Mo over time and explains the deactivation as observed in catalytic tests.
The novel freeze-quench approach, which can trap reaction solutions within 1 s of mixing, opens up a
large field of homogeneous catalysis and liquid chemistry to be studied, being able to quench this rapidly,
whilst characterisation techniques with long data acquisition can be performed.
Keywords:
Oligomerisation
Catalysis
XAS
Mechanism
Mo
Stopped-flow
Freeze-quench
Homogeneous
Ó 2011 Published by Elsevier Inc.
1. Introduction
The extremely high selectivity for 1-hexene is thought to occur
via a mechanism based upon a metallocyclic intermediate formed
The efficient catalytic conversion of small molecules into more
complex species in commercial demand continues to be a very
important feature of the global chemical industry, and the ability
to increase the specificity and selectivity of these processes is
essential for clean, energy efficient processes. Oligomerisation of
alkenes usually occurs via transition metal or aluminium catalysed
processes to produce mixtures of alkenes in the C4–C26 range [1,2].
However, the selective trimerisation and tetramerisation of ethene
to produce the linear alpha-olefins (LAOs) 1-hexene and 1-octene,
respectively, are of major significance due to the importance of
these co-monomers in the production of linear low-density poly-
ethylene (LLDPE) [2–4]. LLDPE accounts for around 50% of the
LAO co-monomers produced industrially. A variety of transition
metal catalysts facilitate the selective trimerisation of ethene, most
of which are based on early transition metals such as titanium, tan-
talum or, most importantly, chromium [2–4]. These form the basis
of several key industrial catalysts, including the Phillips pyrrolide
system, the Sasol mixed heteroatom systems and the BP diphos-
phine systems [5–7].
through the reaction of Cr(III) pre-catalysts with ethene in the pres-
ence of methylaluminoxane (MAO) co-catalyst [2,6,8]. The auxiliary
ligand on Cr(III) is generally a tridentate ligand incorporating Group
15 or 16 donor atoms, e.g. [CrCl3(SNS)] (SNS = RS(CH2)2NH(CH2)2SR)
and [CrCl3{R2P(CH2)2NH(CH2)2PR2}] (R = long-chain alkyl group).
Chromium complexes containing the ligand Ph2PN(iPr)PPh2 (PNP)
are highly selective catalyst precursors for the tetramerisation of
ethylene to 1-octene when activated with MAO, and improved activ-
ities are achieved when aluminate or BARF activators are introduced
[9–11]. The pursuit of catalysts with ever higher specificities
and selectivities for these oligomerisations and polymerisations
demands a detailed understanding of the individual stages of
the catalytic cycle, the dependence upon metal, promoter and/or
co-catalyst. However, developing such catalyst systems is hindered
by characterisation difficulties due to the paramagnetism of the
majority of the Cr complexes (precluding NMR analysis), and since
MAO is not a single chemical entity, therefore its precise role is dif-
ficult to establish. The role of the ligand donor types (hard versus
soft) and architectures are undoubtedly very important in defining
the mechanisms in these processes, promoting different reaction
pathways. Further, whilst Cr complexes with softer P/S donor atoms
tend to lead to trimer/tetramer, the use of harder donor ligands only
(e.g. amines) favours a Shultz–Flory distribution of oligomers [2].
⇑
Corresponding author. Fax: +49 89 28912866.
E-mail address: moniek.tromp@tum.de (M. Tromp).
0021-9517/$ - see front matter Ó 2011 Published by Elsevier Inc.
doi:10.1016/j.jcat.2011.10.015

248
S.A. Bartlett et al. / Journal of Catalysis 284 (2011) 247–258
Work on the industrially important [CrCl3(SNS)] with various
We have pursued two methods to allow easier characterisation
of these catalytic systems: (i) substituting the Cr(III) for the heavier
4d Mo(III) organometallic systems to model the early stages of the
activation and (ii) modification of the stopped-flow system in order
to maintain the advantages of the time resolution of the instru-
ment, but also trapping the intermediates to allow long data acqui-
sition via XAS to be conducted.
alkylating agents by Gambarotta and co-workers suggests that at
low Cr:Al ratios Cl-bridged dimers are formed, including
[{(SNS)CrMe}2(l
-Cl)2]2+ [12]. Braunstein et al. have also recently
shown that for the [CrCl3(NPN)] (NPN = bis(2-picolyl)phenylphos-
phine) chloro-bridged dimers, as well as mononuclear clusters
with a mixture of alkyl and halide ligands can be formed, depend-
ing on the alkylating agent and reaction conditions [13]. It is how-
ever important to note that the ratio of Cr:MMAO in the operating
catalyst system is significantly higher than used in these studies.
Previously, we have examined a range of Cr(III) complexes with
various neutral tridentate N-, S- and O-donor ligands comprised of
1.1. Mo(III) complexes
We have prepared the analogous Mo(III) trichloro complexes,
[MoCl3(L)] (L = SNS and SSS) and investigated their utility as stoi-
chiometric model systems for the Cr catalysts by determining their
reactivities with AlMe3. [MoCl3(SNS-R)] complexes for a variety of
R groups have been reported previously, although they show neg-
ligible catalytic activity towards ethene oligomerisation [21]. How-
ever, it is reasonable that the Mo systems might undergo similar
chemistry in the early stages of activation, the slower kinetics at
Mo will aid identification of individual species in the reaction. Fur-
thermore, the Mo systems are also chosen as model systems be-
cause of the higher Mo K-edge energy at which the X-ray
experiment can be undertaken and thus the increased time resolu-
tion of the XAS experiment to be obtained. Mo will allow time-re-
solved transmission studies in which real reaction intermediates
are likely to be probed instead of the final (often decomposed)
end-states of the Cr systems. With the Mo K-edge being present
in the similar high energy range of Pd (i.e. ꢀ20 keV)16 the potential
of the stopped-flow EDE approach for these systems is high. In a
further modification, the [MoBr3(L)] analogues were also synthes-
ised in order to aid distinction between the S of the ligand and
the halide in the EXAFS analyses, which is not possible for S and
Cl since they are corresponding neighbours in the periodic table
with similar backscattering amplitudes.
three distinct architectures: facultative (S(CH2CH2SC10H21 2
)
(SSS-Decyl) and O(CH2CH2SC10H21)2) (SOS-Decyl), tripodal (MeC
(CH2SC4H9)3, MeC(CH2SC10H21)3) and macrocyclic (C10 H21)3[9]
aneN3 ([9]aneN3 = 1,4,7-triazacyclononane) as catalysts for olefin
oligomerisation and polymerisation and probed the identity of the
species before and after addition of the molecular alkylating agent,
AlMe3, using IR, UV/visible and EPR spectroscopy along with X-ray
Absorption Fine Structure (XAFS) studies [14]. Treatment of these
complexes with excess (300 mol. equivs.) modified MAO (MMAO)
leadstocatalyticallyactivespeciesandshowssignificantdependence
upon the ligand donor set. UV/visible and EPR measurements suggest
that a change in the oxidation state of the chromium metal centre oc-
cursinalltheseexamplesuponadditionofexcessAlMe3, lendingsup-
port to the proposal that the catalysis involves a Cr(II)/Cr(IV) redox
cycle. The partial structures of the AlMe3-activated species were
probed via XAFS spectroscopy. It appears that when the Cr(III) com-
plexes are treated with AlMe3, the Cl ligands are substituted by
methyl groups and some ligands substantially dissociate from the
metal centre; with the resulting species catalysing the oligomeri-
sation and polymerisation of ethylene.
Whilst these are the first XAFS studies probing the activation of
the Cr-based oligomerisation/polymerisation pre-catalysts, the
timescale for the measurements (usually 2–4 h to acquire suffi-
cient data from ꢀ5 mM solutions – corresponding to the approxi-
mate Cr concentration in the operating catalyst – for partial
structure analysis) mean that, by definition, these are ‘end-state’
investigations. The information that can be obtained from such
measurements provides limited insight into the active species gi-
ven that under the operating conditions the catalyst lifetime is typ-
ically <1 h.
XAS (in energy dispersive mode (EDE)) was for the first time
successfully synchronised with a stopped-flow experiment in
1990, in which the oxidation of FeIII aq by hydroquinone was inves-
tigated [15]. More recently, stopped-flow systems have been
developed for homogeneous catalysis studies, to probe homoge-
neous catalytic intermediates in situ and time-resolved (millisec-
onds) [16–18]. In the stopped-flow system, the evolution of the
reaction mixture can be monitored in time with EDE or Quick-
XAS (QEXAFS), allowing full Extended X-ray Absorption Fine Struc-
ture (EXAFS) spectra to be obtained in the millisecond to second
timescale [19]. The problem with this set-up is that XAFS data
acquisition in fluorescence mode is not possible (only transmis-
sion), and therefore, a relatively high concentration of the species
under investigation is required. At the same time, the complete
reaction mixture, i.e. including solvents and reagents, should not
be too absorbing at the X-ray energies required for the experiment.
The stopped-flow EDE method has been proven for homogeneous
Pd16 and Re [20] systems with a high energy Pd K- and Re L-edge,
enabling the in situ transmission experiments. However, although
the method has been proven for homogeneous Ni (8.333 keV)16
and Cu (8.979 keV)18 systems, the Cr K-edge is at an even lower en-
ergy of 5.989 keV. Not only the solvent and reactant molecules, but
even a few cm of air, are highly absorbing at these energies, se-
verely hampering the transmission experiment.
1.2. Stopped-flow freeze-quench XAS development
In order to overcome the inherent obstacles associated with
stopped-flow XAS studies, as mentioned, and to allow accurate
characterisation of reaction intermediates, a XAFS cell that allows
‘trapping’ (stabilisation) of the intermediate species present at var-
ious stages of activation and catalysis reaction, by rapid quenching
of the catalyst system is required. With the catalytic intermediate
‘trapped’, XAFS experimentation can subsequently be performed in
fluorescence mode with long data acquisition times as required for
low energy systems like Cr. The development of such a XAFS
quench cell to allow highly reactive transients to be trapped on a
sub-second timescale and analysed structurally and electronically
will be complementary to the established stopped-flow EDE proce-
dure, allowing in principle the in-depth study of any homogeneous
process. This approach will provide detailed structural and elec-
tronic insights in homogeneous (catalytic) reaction mechanisms.
Although freeze-quench (also in combination with stopped-flow
methodologies) is used frequently for biological and bio-inorganic
systems [22], this is a new approach in homogeneous catalysis and
requires considerable modifications of the apparatus (vide infra).
Here, we report on the Mo(III) organometallic complexes and
their activation with AlMe3, as well as a special quench-freeze
XAFS cell designed to allow detailed XAFS and XANES studies to
be performed to probe the early stages of activation and catalysis
and thus provide a more detailed understanding of the activation
stages of the polymerisation and oligomerisation catalyst systems
rather than end-state studies. The results are shown for Mo sys-
tems, and their analogy to the Cr is made.

S.A. Bartlett et al. / Journal of Catalysis 284 (2011) 247–258
249
3158 (NH), 324br sh), 311s (Mo–Cl). Micro-analysis required for
24H51Cl3MoNS2: C = 46.5; H = 8.3; N = 2.3; found C = 46.6;
2. Experimental
C
H = 8.4; N = 2.5%.
2.1. Mo complex synthesis and characterisation
2.1.4. [MoCl3{S(CH2CH2SCH2–C6H4-p-C(CH3)3)2}] [MoCl3(SBz)]
S(CH2CH2SCH2–C6H4-p-C(CH3)3)2 (0.59 g, 1.31 mmol) was added
directly to a suspension of [MoCl3(THF)3] (0.5 g, 1.19 mmol) in THF
(20 mL). The solution was allowed to stir for 3 d. under N2 to give a
very dark solution. The solvent was then removed in vacuo to give
a very dark solid. Dry hexane (25 mL) was added and the solution
was allowed to stir for 30 min. The red solid was then filtered under
N2 and dried in vacuo. Yield: 0.56 g, 0.86 mmol (72%). UV/Vis spec-
All complexes were synthesised under inert conditions using
standard Schlenk line techniques, using dry solvents, having previ-
ously been distilled from appropriate drying agents. Complexes
[MoCl3(THF)3] [23] and [MoBr3(THF)3] [24] and ligand
HN(CH2CH2SC10H21)2 [25] were synthesised in accordance with
previously published procedures. All metal complexes were stored
under inert atmosphere in a glove box. AlMe3 was purchased from
Aldrich (2.0 M in hexanes) and titrated using the published proce-
dure [26]. All solvents used in the stopped-flow instrument were
purchased (anhydrous) from Aldrich and used as received.
trum (toluene): kmax 476 nm,
e
= 938 molꢂ1 dm3 cmꢂ1. IR (Nujol
mull, cmꢂ1): 367 m. 352sh, 322s (Mo–Cl). Micro-analysis required
for C26H38Cl3MoS3: C = 48.1; H = 5.9; found C = 47.8; H = 5.8%.
1H and 13C{1H} NMR spectra were recorded on a Bruker AV300
spectrometer. Micro-analyses were obtained through Medac Ltd.,
Egham UK. Infrared spectra were recorded as a Nujol mull between
CsI plates under inert conditions, using Perkin Elmer FT-IR Spec-
trum 100 Spectrometer across the range 4000–200 cm–1. Only
the key bands associated with the NH function and the Mo–Cl
stretching vibrations are quoted below. UV–visible spectra were
recorded both using the stopped-flow instrument (Fibre optic J &
M Analytic MCS-UVNIR) and using a Perkin Elmer Lambda 19 spec-
trometer either in solution in a 1-cm path-length quartz cuvette, or
as solids by diffuse reflectance (DR). All spectroscopic samples
were prepared freshly in a dry N2-purged glove box.
2.1.5. [MoBr3{HN(CH2CH2SC10H21)2}] [MoBr3(SNS-Decyl)]
Prepared in the same manner as [MoCl3(SNS-tBu)] above, using
[MoBr3(THF)3] (0.6 g, 1.09 mmol) and HN(CH2CH2SC10H21 2
(0.55 g, 1.31 mmol), with 1.5 d. stirring after the addition of CH2Cl2.
UV/Vis spectrum (CH2Cl2): kmax 486 nm (sh), = 1000; 421 nm,
= 1625 molꢂ1 dm3 cmꢂ1 IR (Nujol mull, cmꢂ1): 3172 (NH),
257 m, 243sh, 228 m (Mo–Br). Micro-analysis required for
24H51Br3MoNS2: C = 38.3; H = 6.8; N = 1.9; found C = 39.0;
)
e
e
.
C
H = 7.2; N = 2.0%.
2.1.6. [MoBr3{S(CH2CH2SCH2–C6H4-p-C(CH3)3)2}] [MoBr3(SBz)]
S(CH2CH2SCH2–C6H4-p-C(CH3)3)2 (0.49 g, 1.09 mmol) was added
directly to a suspension of [MoBr3(THF)3] (0.5 g, 0.91 mmol) in THF
(10 mL). The solution was allowed to stir for 1 d. under N2 to give an
orange precipitate. The solvent was then removed in vacuo and
CH2Cl2 added to the orange solid to ensure all starting material
had dissolved. This was allowed to stir overnight and then filtered
under N2. The orange solid was then washed with further CH2Cl2, fil-
tered under N2 and dried in vacuo. Yield: 0.38 g, 0.49 mmol (54%).
2.1.1. S(CH2CH2SCH2–C6H4-p-C(CH3)3)2 [SBz]
Sodium (2.67 g, 116.3 mmol) was slowly dissolved in a solution
of ethanol (200 mL) and bis(2-mercaptoethyl)sulphide (7.16 g,
46.40 mmol) under an N2 atmosphere The solution was heated to
reflux and turned cloudy upon the dropwise addition of 4-tert-
butylbenzylbromide (21.36 g, 94.03 mmol); this was then refluxed
overnight under N2. Ethanol was removed in vacuo, and the reac-
tion was hydrolysed using water (50 mL). After extraction with
diethyl ether (3 ꢁ 30 mL), the organics were combined, dried over
MgSO4, filtered and the solvent removed in vacuo. The yellow oil
was further dried under vacuum. Yield: 18.15 g, 40.6 mmol (88%).
1H NMR – ppm (CD2Cl2): 1.32 (s, [18H], CH3), 2.63 (m, [8H],
CH2S), 3.72 (s, [4H], C6H4–CH2S), 7.2, 7.4 (dd, [8H], Ar). 13C{1H}
NMR – ppm (CD2Cl2): 31.7 (CH3), 32.1, 32.5 (CH2S), 36.4 (C6H4–
CH2S), 126.0, 129.0 (ArCH), 135.8, 150.7 (ArC).
UV/Vis spectrum (toluene): kmax 472 nm, e .
= 2300 molꢂ1 dm3 cmꢂ1
IR (Nujol mull, cmꢂ1): 287 m, 273 m, 265 m (Mo–Br). Micro-analysis
required for C26H38Br3MoS3: C = 39.9; H = 4.9; found C = 40.1;
H = 4.8%.
2.1.7. [MoCl3([9]aneS3)] ([9]aneS3 = 1,4,7-trithiacyclononane)
Prepared by a modification of the literature method [27].
[9]aneS3 (0.1 g, 0.6 mmol) in THF (5 mL) was added to [MoCl3(THF)3]
(0.2 g, 0.6 mmol) in THF (5 mL). This was allowed to stir for 2 h at
room temperature to give a cloudy red solution. The solvent was
then reduced to ꢀ2 mL in vacuo, and CH2Cl2 (20 mL) added. The reac-
tion was left to stir overnight. The red precipitate was then filtered,
washed with CH2Cl2 (2 ꢁ 5 mL) and dried in vacuo to give an orange
solid. Yield: 0.10 g, 0.26 mmol (48%). DR UV/Vis spectrum (BaSO4):
480 nm. IR (Nujol mull, cmꢂ1): 315s, 292 m (Mo–Cl). Micro-analysis
required for C6H12Cl3MoS3: C = 18.8; H = 3.2; found C = 19.1;
H = 3.1%.
2.1.2. [MoCl3{HN(CH2CH2SC(CH3)3)2}] [MoCl3(SNS-tBu)]
HN(CH2CH2SC(CH3)3)2 (0.9 g, 3.6 mmol) was added directly to a
suspension of [MoCl3(THF)3] (1.0 g, 2.4 mmol) in THF (15 mL), giv-
ing an orange solution after 2 h. This was then allowed to stir over-
night under N2. The solution was filtered under N2 and washed
with cold CH2Cl2 (3 ꢁ 5 mL). The yellow solid was dried in vacuo.
Yield: 0.7 g, 1.6 mmol (64%). DR UV/Vis Spectrum (CaF2): kmax
446 nm. IR (Nujol mull, cmꢂ1): 3144 (NH), 336s, 313s, 303s (Mo–
Cl). Micro-analysis required for
C12H27Cl3MoNS2: C = 31.9;
H = 6.0; N = 3.1; found C = 32.4; H = 6.1; N = 2.6%.
2.2. X-Ray crystallography
2.1.3. [MoCl3{HN(CH2CH2SC10H21)2}] [MoCl3(SNS-Decyl)]
Details of the crystallographic data collection and refinement
parameters are given in Table 1. Crystals of [MoCl3([9]aneS3)]
and weakly diffracting crystals of [MoCl3{tBuS(CH2)2NH(CH2)2St-
Bu}] suitable for single crystal X-ray analysis were obtained from
a solution of the complex in THF or CH2Cl2 solution (at ꢂ18 °C),
respectively. Data collections used a Bruker–Nonius Kappa CCD
HN(CH2CH2SC10H21)2)2 (1.4 g, 3.6 mmol) was added directly to
a suspension of [MoCl3(THF)3] (1.0 g, 2.4 mmol) in THF (20 mL).
The mixture was allowed to stir overnight under N2, giving a very
dark solution. The THF was then reduced to 5 mL in vacuo, where
CH2Cl2 (15 mL) was added and the mixture was allowed to stir
for a further 14 h under N2. All the solvent was then removed in va-
cuo, and dry hexane added. The solution was stirred and sonicated,
and the dark brown solid was then filtered under N2 and dried in
vacuo. Yield: 1.27 g, 2.1 mmol (86%). UV/Vis spectrum (toluene):
diffractometer fitted with monochromated Mo
Ka radiation
(k = 0.71073 Å), with the crystals held at 120 K in a nitrogen gas
stream. Structure solution and refinement were straightforward
[28]. H atoms bonded to C were introduced into the models in
idealised positions using the default C–H distance. The crystal
kmax 455 nm,
e .
= 350 molꢂ1 dm3 cmꢂ1 IR (Nujol mull, cmꢂ1):

250
S.A. Bartlett et al. / Journal of Catalysis 284 (2011) 247–258
Table 1
Crystal data and structure refinement detailsa.
Compound
[MoCl3([9]aneS3)]
[MoCl3{tBuS(CH2)2NH(CH2)2StBu}]
Formula
M
Crystal system
Space group (no.)
a (Å)
C6H12Cl3MoS3
382.63
Tetragonal
I41cd (110)
17.213(2)
17.213(2)
16.389(2)
90
C12H27Cl3MoNS2
451.76
Orthorhombic
P212121 (19)
6.890(2)
9.349(3)
30.338(9)
90
b (Å)
c (Å)
a
(°)
b (°)
90
90
c
(°)
90
90
U (Å3)
4855.7(10)
16
2.210
1954.3(10)
4
1.284
Z
l
(Mo K
a
)/mmꢂ1
Total no. reflections
Unique reflections
Rint
16,285
2579
0.0938
118, 1
1.116
22,151
3826
0.170
173, 24
1.112
0.101
No. of parameters, restraints
Goodness-of-fit on F2
b
R1 [Io > 2
r(Io)]
0.050
R1 (all data)
0.069
0.129
wR2b [Io > 2
wR2 (all data)
r
(Io)]
0.089
0.097
0.2219
0.237
a
Common items: temperature = 120 K; wavelength (Mo K
a) = 0.71073 Å; h(max) = 27.5°.
2
1=2
R1
=
R
||Fo| ꢂ |Fc||/
R
|Fo|. wR2 ¼ ½
R
wðF2o ꢂ F2c Þ =
R
wF4oꢃ
.
b
quality for [MoCl3{tBuS(CH2)2NH(CH2)2StBu}] was modest, and
hence structure determination is of modest quality, although the
core geometry and isomer determination are not in doubt, detailed
comparisons of the geometric parameters should be treated with
care. The structures are shown in Figs. 1 and 2. Selected bond
lengths and angles are presented in Tables 2 and 3. CIF files of
the crystal structure data are available as Electronic Supplemen-
tary Data files.
ing of individual stages of the cycle(s). The stopped-flow system
has a proven performance in controlled and reproducible injection
of accurate mixtures in millisecond timescales. Therefore, the set-
up as developed here is based on the stopped-flow EDE approach,
using a commercially available stopped-flow system with a proven
performance for combined EDE/UV–Vis experimentation [16,18].
The stopped-flow (SFM400/QS, Biologic, France) consists of four
syringes that can be filled with reaction solution or solvent [18,29].
The system is computer controlled and allows injection of very
precise volumes with controlled injection rates (and thus injection
times). The solutions are injected using stepper motors via the de-
lay lines and the mixers into a cuvette or capillary, depending on
the requirement. Specially designed quartz cuvettes allow time-re-
solved EDE experiments to be performed on the injected mixtures.
Using cuvettes with two transmission pathways perpendicular to
each other will allow the addition of a complementary technique,
like UV/Vis, to be performed simultaneously [18]. An optical fibre
UV/Vis spectrometer (MCS–UV–VIS–NIR 1/500-3 Fast Diode Array
Detector, BioLogic, France), specially customised for the stopped-
flow, is also used in this study. The UV–Vis technique is used to
perform time-resolved UV/Vis studies (in the laboratory) as well
as combined EDE/UV/Vis experiments (at the synchrotron). Due
to the air and moisture sensitive nature of the chemistry, the entire
stopped-flow is connected to a Schlenk line allowing us to work
under an argon atmosphere. To ensure the stopped-flow is com-
pletely air and moisture free, a careful cleaning and passivating
procedure is employed before every experiment. The internal com-
ponents of the stopped-flow instrument were manufactured in
KelFÒ (polychlorotrifluorethyene), with KalrezÒ (perfluoroelastom-
er) rings, to ensure full chemical compatibility.
2.3. XAFS cell development
We have developed a freeze-quench (FQ) XAFS cell to permit
measurements to be made following rapid quenching of the cata-
lyst mixture. This is essential in order to be able to probe the early
stages of activation and catalysis in the true trimerisation and tetr-
amerisation mixtures and to provide a more detailed understand-
The standard freeze-quench accessory as available from Bio-
Logic was developed for non-sensitive water solution, spraying
through air into isopentane. Because of the air and moisture sensi-
tivity or our chemistry, this was not a viable option. Therefore, we
have developed a XAFS measurement cell with freeze-quench sys-
tem to be attached to the stopped-flow, Fig. 3. A KaptonÒ capillary
is mounted into a custom holder attached to the standard BioLogic
freeze-quench accessory. The holder allows direct injection into
the KaptonÒ tube whilst an argon gas is flowed around the spray
exit and tube inlet and prevents ingress of air (or moisture) into
Fig. 1. View of the structure of [MoCl3([9]aneS3)] with number scheme adopted. H
atoms are omitted for clarity and ellipsoids are drawn at the 50% probability level.

S.A. Bartlett et al. / Journal of Catalysis 284 (2011) 247–258
251
Fig. 2. View of the structure of [MoCl3{tBuS(CH2)2NH(CH2)2StBu}] with number scheme adopted. H atoms are omitted for clarity and ellipsoids are drawn at the 50%
probability level.
Table 2
Selected bond lengths (Å) and angles (°) for [MoCl3([9]aneS3)].
2.4. XAFS experimentation
Mo1–Cl1
Mo1–Cl3
Mo1–S2
2.402(2)
2.385(2)
2.520(2)
96.78(9)
96.57(9)
172.42(9)
86.12(7)
169.58(8)
168.16(7)
89.59(7)
83.76(6)
Mo1–Cl2
Mo1–S1
Mo1–S3
Cl2–Mo1–Cl1
Cl2–Mo1–S1
Cl1–Mo1–S1
Cl3–Mo1–S3
S1–Mo1–S3
Cl3–Mo1–S2
S1–Mo1–S2
2.345(2)
2.464(2)
2.484(2)
99.53(8)
88.60(8)
87.77(7)
91.41(9)
83.60(7)
89.60(9)
84.20(8)
All samples for solid state Mo K-edge EXAFS analysis were mixed
with boron nitride, pressed into pellets under inert atmosphere and
encapsulated in KaptonÒ tape. All solution data collected for Mo K-
edge EXAFS were obtained from solutions made up to the stated con-
centrations with anhydrous toluene under inert conditions. The
solution was then transferred to a dry, argon-purged and sealed li-
quid XAFS cell [14] for analysis on the beamline. Reactions were also
performed in this liquid XAFS cell by adding of the stated amount of
AlMe3 (2.0 M in hexane) (using a syringe) directly into the XAFS cell.
The solution was then analysed with Mo K-edge EXAFS after ca.
5 min of reaction. In the rest of this paper, this cell is referred to as
‘large solution cell’ (in which we have measured reaction ‘start-
and end-states’).
Cl2–Mo1–Cl3
Cl3–Mo1–Cl1
Cl3–Mo1–S1
Cl2–Mo1–S3
Cl1–Mo1–S3
Cl2–Mo1–S2
Cl1–Mo1–S2
S2–Mo1–S3
Table 3
Selected bond lengths (Å) and angles (°) for [MoCl3{tBuS(CH2)2NH(CH2)2StBu}].
All stopped-flow and freeze-quench experiments were carried
out using a BioLogic SFM-400 stopped-flow instrument as ex-
plained above. In stopped-flow mode, all reactions were observed
Mo1–N1
Mo1–Cl2
Mo1–S1
2.208(12)
2.410(4)
2.521(4)
85.2(3)
91.28(13)
173.66(14)
80.0(3)
101.14(13)
81.2(3)
98.46(13)
Mo1
Mo1–Cl1
Mo1–Cl3
Mo1–S2
2.384(4)
2.452(4)
2.532(4)
176.4(3)
88.5(3)
95.01(13)
95.90(13)
82.06(12)
96.92(13)
83.02(12)
156.26(12)
N1–Mo1–Cl1
Cl1–Mo1–Cl2
Cl1–Mo1–Cl3
N1–Mo1–S1
Cl2–Mo1–S1
N1–Mo1–S2
Cl2–Mo1–S2
S1
N1–Mo1–Cl2
N1–Mo1–Cl3
Cl2–Mo1–Cl3
Cl1–Mo1–S1
Cl3–Mo1–S1
Cl1–Mo1–S2
Cl3–Mo1–S2
S2
in cuvette TC-100/10T with minimum dead volume of 30.2 lL. In
time UV/Vis was collected using J and M Analytic AG MCS-UVNIR
500-3 fibre optic diode array spectrometer. When using the
freeze-quench attachment, the same set-up is used (with UV–Vis
and cuvette head removed) giving a total minimum dead volume
of 19 lL from mixing to ejection, freezing time in toluene less than
1 s. Prior to all reactions using the stopped-flow instrument, the
whole instrument was purged with argon, maintaining a positive
pressure throughout (including waste lines) by attachment to a
Schlenk line. All lines were washed with anhydrous toluene, then
the subsequent reactant, followed by further washing with anhy-
drous toluene.
the sample. The KaptonÒ tube is sealed with a plug, and immersed
into liquid nitrogen in a Dewar (before and during injection). As
such the reaction mixture can be injected (collected) and frozen di-
rectly and maintained at 77 K. The injection and freezing time was
estimated to be below 1 s. Different delays in the stopped-flow al-
low the accurate timing of reaction intermediates. For XAS exper-
imentation, the tube with frozen sample is subsequently placed
into a holder in the X-ray beam, whilst being cooled continuously
using a Cryostream (100 K). By setting up the appropriate cryo flow
rate, ice formation on the outside of the capillary is restricted (this
is not crucial for the high energy Mo XAS data as obtained in this
study, but will be when looking at low energy systems like the
3d transition metals with much lower X-ray edge energies).
The total set-up thus consists of different parts that can be used
independently or as an ensemble. As such, the system is very
versatile and allows a broad range of experiments to be performed;
reactions at variable temperatures, direct time-resolved studies or
via quench-freeze methods (from any given reaction temperature)
allowing long data acquisition times (also non-XAS) if required.
The Mo K-edge XAFS measurements were performed at differ-
ent synchrotrons across Europe. Solid and solution (large solution
cell) preliminary experiments were performed at BM26 (DUBBLE)
[30] of the European Synchrotron Radiation Facility (ESRF) in Gre-
noble, France. Mo K-edge XAS data were obtained using a Si(111)
double crystal monochromator using ionisation chambers for
transmission detection, with acquisition times of 30 min (three
scans were averaged to improve S/N unless stated otherwise). Solid
and solution (large cell) experiments, as well as time-resolved
stopped-flow QEXAFS/UV/Vis were performed at the SuperXAS
beamline at the Swiss Light Source in Villigen, Switzerland. The
time-resolved Mo K-edge QEXAFS experiments were performed
using a fast scanning Si(111) double crystal monochromator, with
a time resolution of 1 min/spectrum for the stopped-flow and
3min/spectrum for the solids and large solution cells. All data from
these experiments were obtained in transmission mode using

252
S.A. Bartlett et al. / Journal of Catalysis 284 (2011) 247–258
Fig. 3. Stopped-flow Freeze-Quench accessory: A Perspex box is constructed to be mounted to the standard spray outlet of the stopped-flow quench accessory and to hold the
KaptonÒ tube. A gas inlet and outlet provide an argon gas atmosphere from the spray exit to the tube inlet. The right picture shows the KaptonÒ tube, containing frozen Mo
intermediate, positioned in the X-ray beam using a red laser whilst in the cryostream.
ionisation chambers. Time-resolved QEXAFS/UV–Vis as well as
freeze-quench experiments were performed at the B18 beamline
[31] at Diamond Light Source in Didcot, England. A Si(111) double
crystal was used in combination with ion chamber detectors for
time-resolved transmission data and an OrtecÒ Ge 9 element Solid
State detector for fluorescence acquisition on the frozen samples.
Mo K-edge step-scans were obtained in 35 min, whereas the QEX-
AFS data were obtained with a time resolution of 30 s per spec-
trum. The time-resolved spectra shown in this paper are a
combination of data obtained at the SLS and Diamond. All spectra
were calibrated using a Mo foil. XAS data processing and EXAFS
analysis were performed using IFEFFIT [32] with the Horae package
[33] (Athena and Artemis). The amplitude, i.e. s20, was derived from
EXAFS data analysis of known Mo reference compounds (with
known coordination numbers which were fixed during analysis)
to be 0.85, which was used as a fixed input parameter in all fits
to allow coordination number (CN) refinement. The organometallic
complexes were fitted using single crystal models or similar com-
plexes, as indicated. The detailed fitting parameters are given in
the results section.
The solid and solution structures of the complexes were con-
firmed by EXAFS and compared to the other characterisation ob-
tained. The Mo K-edge EXAFS results of the chloride complexes
[MoCl3(SNS-Decyl)] and [MoCl3(SBz)] are presented in Tables 4
and 5, and the Mo K-edge EXAFS results on the bromide analogue
[MoBr3(SBz)] in Table 6. Good fits have been obtained for all data
sets, with low R-factors. Representative EXAFS data and fits are
presented in Fig. 4, for [MoBr3(SBz)] solid and solution. Some
parameters with higher errors or Debye Waller factors are ob-
tained and discussed when they appear.
The mer-[MoCl3(SNS-Decyl)] complex was characterised by
three chlorine atoms at 2.41(3) Å, two sulphurs at 2.50(2) Å and
one nitrogen atom at ꢀ2.1(1) Å. The crystal structure data of the
tBu-analogue, mer-[MoCl3(SNS-tBu)] (Tables 1 and 3), were used
as an input model for the fit. The contribution of the nitrogen atom
directly bonded to the Mo is very weak and difficult to identify,
with a large error in distance and a large Debye Waller factor,
due to the presence of the 5 higher Z sulphur and chlorine atoms.
The contribution can be omitted and a good fit is obtained, how-
ever, crystal structure data (Mo–N 2.208(12) Å based on XRD –
The XANES data are not discussed in this paper. Because of the
focus of the experiment onto obtaining time-resolved EXAFS at the
best quality possible, the XANES data were obtained at relatively
low resolution. Moreover, the changes in the XANES of the Mo K-
edge are very small because of the broad edge at these high X-
ray energies (long core hole lifetimes), which makes the differences
between Mo3+, and potentially formed Mo2+ or Mo4+ very difficult
to observe, especially since the K-edge reflects the dipole allowed
1s? 5p transition and only probes the 4d states indirectly.
see Table 3) on the [MoCl3(SNS-tBu)] and IR results (via mNH) con-
firm its presence and thus a fixed coordination of one Mo–N is in-
cluded in the fits. Because of the proximity of chlorine and sulphur
in the periodic table, they cannot be distinguished based on EXAFS
only. Their different assignment here was based on chemical
knowledge on these compounds. The coordination numbers were
fixed to simplify the fit and allow us to distinguish between the
two shells.
The Mo–Cl distance of 2.41 Å corresponds well with an average
of the three distances as observed in the [MoCl3(SNS-tBu)] crystal
3. Results and discussion
Table 4
Mo K-edge EXAFS data analyses results on MoCl3(SNS-Decyl) complex (all fitted in
combined k1–3-weighting).
3.1. Solid state and solution complexes
Abs Sc
CN
R (Å)
2
r
2(Åꢂ2
)
Fitting parameters, k (Åꢂ1
)
A series of Mo(III) complexes was successfully synthesised and
characterised with standard techniques like IR and UV–Vis spectros-
copy and micro-analysis (see Section 2): [MoCl3 {HN (CH2 CH2
SC10H21)2}] [MoCl3(SNS-Decyl)], [MoCl3(SNS-tBu)], [MoCl3 {S
(CH2CH2SCH2–C6H4-p-C(CH3)3)2}] [MoCl3(SBz)], [MoBr3 {S (CH2
CH2SCH2–C6H4-p-C(CH3)3)2}] [MoBr3(SBz)] and [MoCl3([9]aneS3)];
[9]aneS3 = 1,4,7-trithiacyclononane. Only for the [MoCl3(SNS-t
Bu)] and the [MoCl3([9]aneS3)] complexes were crystal structures
obtained, so the structures of all complexes in the solid state and
solution were carefully analysed to confirm their integrity and pro-
vide good starting points for the reaction studied.
Solid [MoCl3(SNS-Decyl)]
Mo–N
Mo–Cl(S)
Mo–S(Cl)
1(fix)
3(fix)
2(fix)
2.1(1)
2.41(3)
2.50(2)
0.010(6)
0.002(3)
0.001(2)
2.0 < k < 14.7; 1.1 < R < 2.57
Amp = 0.85; E0 = 3(1); R = 0.01
Solution [MoCl3(SNS-Decyl)]
Mo–N
Mo–Cl(S)
Mo–S(Cl)
1(fix)
3(fix)
2(fix)
2.1(1)
2.40(2)
2.53(2)
0.015(9)
0.002(1)
0.002(2)
2.0 < k < 15.7; 1.1 < R < 2.7
Amp = 0.85; E0 = 3(1); R = 0.02
[MoCl3(SNS-Decyl)] + 20 equivalent AlMe3 ꢂ large solution cell
Mo–N/C
Mo–S(Cl)
3(1)
1.9(9)
2.22(9)
2.50(3)
0.010(9)
0.003(4)
2.0 < k < 11; 1 < R < 3
Amp = 0.85; E0 = 3(3); R = 0.05

S.A. Bartlett et al. / Journal of Catalysis 284 (2011) 247–258
253
Table 5
in Table 4, the results are similar to the solid state structure, as ex-
pected. Good fits have been obtained for solid state and solution,
with low R-factors.
Mo K-edge EXAFS data analyses results on MoCl3(SBz) complex (all fitted in combined
k
1–3-weighting).
Abs Sc
CN
R (Å)
2
r2 (Åꢂ2
)
Fitting parameters, k (Åꢂ1
)
The solid state structure for the [MoCl3(SBz)] complex was
characterised with Mo K-edge EXAFS and its results presented in
Table 5. The crystal structure of the [MoCl3([9]aneS3)] (see Tables
1 and 2) model compound was used as an input model for the fit-
ting procedure. The structure of the [MoCl3(SBz)] complex could
be refined with three chlorine atoms at 2.40(2) Å and three sulphur
at 2.43(2) Å, with low Debye Waller factors. The two shell analysis
is consistent with a facial geometry as observed in the single crys-
tal XRD for the [MoBr3(SBz)] (vide infra). The obtained distances
are in the range expected Mo–Cl 2.345(2), 2.385(2), 2.402(2) and
Mo–S 2.464(2), 2.4838(19), 2.520(2) Å for [MoCl3([9]aneS3)]).
Again, the structure in solution is similar to the solid state struc-
ture, as expected.
Table 6 presents the Mo K-edge EXAFS analyses results of the
bromide analogue, i.e. [MoBr3(SBz)]. Two shells can be refined
for the solid state structure; one Mo–Br shell with a coordination
number 2.5(9) and a distance of 2.54(2) Å and a low Debye Waller
factor, and a second shell with 3.5(9) sulphur atoms at 2.45(6) Å
with a large Debye Waller factor (0.010(6)). Whilst crystallo-
Solid [MoCl3(SBz)]
Mo–Cl(S)
Mo–S(Cl)
2.8(3)
2.8(3)
2.40(2)
2.43(2)
0.002(2)
0.003(3)
2.0 < k < 16.8; 1 < R < 2.9
Amp = 0.85; E0 = 1(1); R = 0.01
Solution [MoCl3(SBz)]
Mo–Cl(S)
Mo–S(Cl)
2.8(3)
2.8(3)
2.39(2)
2.46(2)
0.001(2)
0.003(3)
2.95 < k < 16.5; 1 < R < 3
Amp = 0.85; E0 = 1(1); R = 0.02
[MoCl3(SBz)] + 8 equivalent AlMe3 ꢂ stopped-flow ꢀ4 min
Mo–S(Cl)
Mo–C
Mo–Mo
Mo–Mo
Mo–Mo
2.2(9)
3.6(9)
1.3(8)
2(1)
2.55(1)
2.26(1)
2.97(3)
4.36(5)
5.18(7)
0.002(1)
0.002(1)
0.010(5)
0.011(4)
0.015(7)
2.0 < k < 14.6; 1 < R < 5.5
Amp = 0.85; E0 = 6(1); R = 0.01
11(9)
[MoCl3(SBz)] + 20 equivalent AlMe3 ꢂ large solution cell
Mo–S(Cl)
Mo–C
2.5(5)
2.6(4)
2.55(1)
2.23(1)
0.005(2)
0.001(1)
2.0 < k < 15.3; 1.1 < R < 2.67
Amp = 0.85; E0 = 8(1); R = 0.03
structure (Mo–Cl 2.384(4), 2.410(4), 2.452(4)). The Mo–S distance
as detected corresponds well to XRD (Mo–S 2.521(4), 2.532(4)).
The [MoCl3(SNS-Decyl)] complex was subsequently dissolved in
toluene and characterised with Mo K-edge EXAFS. As can be seen
graphic data collected on
a poorly diffracting crystal of
[MoBr3(SBz)] were not of sufficient quality for an accurate structure
Table 6
Mo K-edge EXAFS data analyses results on MoBr3(SBz) complex (all fitted in combined k1–3-weighting).
Abs Sc
CN
R (Å)
2
r2 (Åꢂ2
)
Fitting parameters, k (Åꢂ1
)
Solid [MoBr3(SBz)]
Mo–S
Mo–Br
3.5(9)
2.5(9)
2.45(6)
2.54(2)
0.010(6)
0.002(3)
1.7 < k < 14.2; 1.2 < R < 2.6
Amp = 0.85; E0 = ꢂ4(1); R = 0.007
Solution [MoBr3(SBz)]
Mo–S
Mo–Br
3(fix)
3(fix)
2.45(2)
2.55(1)
0.003(1)
0.006(1)
1.8 < k < 13.5; 1.1 < R < 3
Amp = 0.85; E0 = ꢂ4(1); R = 0.01
[MoBr3(SBz)] + 20 equivalent AlMe3 ꢂ stopped-flow ꢂ after ꢀ4 min
Mo–S
2.5(3)
1.4(9)
2.45(1)
2.59(1)
0.003(1)
0.006(4)
2 < k < 17; 1.1 < R < 3
Amp = 085; E0 = ꢂ4(1); R = 0.01
Mo–Br
[MoBr3(SBz)] + 20 equivalent AlMe3 ꢂ stopped-flow ꢂ freeze-quench ꢂ after ꢀ5 s
Mo–S
2.9(4)
2.4(7)
2.41(1)
2.58(1)
0.004(3)
0.005(3)
2 < k < 14; 1 < R < 3
Amp = 0.85; E0 = ꢂ5(1), R = 0.01
Mo–Br
[MoBr3(SBz)] + 20 equivalent AlMe3 ꢂ stopped-flow ꢂ freeze-quench ꢂ after ꢀ5 min
Mo–S
3.3(3)
1.6(8)
2.38(1)
2.61(1)
0.005(2)
0.005(3)
2 < k < 14; 1 < R < 3
Amp = 0.85; E0 = ꢂ5(1), R = 0.01
Mo–Br
[MoBr3(SBz)] + 20 equivalent AlMe3 ꢂ large solution cell
Mo–S
2.8(5)
0.8(1)
2.43(2)
2.61(2)
0.003(2)
0.003(7)
2 < k < 13; 1 < R < 3
Amp = 0.85; E0 = ꢂ4(2), R = 0.01
Mo–Br
Fig. 4. Mo K-edge k2-weighted EXAFS and Fourier transform data (solid lines) and fits (dotted lines) for [MoBr3(SBz)] solid (black) and solution (red).

254
S.A. Bartlett et al. / Journal of Catalysis 284 (2011) 247–258
determination, it did establish that the complex adopts the
fac-isomer. The fac assignment is also consistent with observations
on Cr(III) and other transition metal complexes with ligands con-
taining –SCH2CH2SCH2CH2S– linkages [34]. With Br being a stron-
ger scatterer the shell with three Mo–Br contributions strongly
dictates the overall fit. The Mo–S shell is less well defined, indicat-
ing a possible distribution of Mo–S distances in the solid state
structure, despite the facial geometry. The solution structure, how-
ever, seems much better defined with a similar refined structure,
but now with low Debye Waller factors for both shells. Because
of the somewhat lower data quality, the coordination numbers
were fixed input parameters.
With the complex solutions being the starting points associated
with activation of the complexes for any of the reactions to be dis-
cussed in the next part, either in the large solution cell or the
stopped-flow, the solution EXAFS data at the start of any reaction
in the cell or the stopped-flow cuvette were obtained and analysed.
All solution structures were identical, independent of the observa-
tion cell.
Å, indicating the neutral ligand remains intact and associated to
the metal (40 min of reaction), whereas the chlorine is replaced
by carbon (i.e. methyl groups) (see Fig. 5). In this case, a very good
fit with low error and Debye Waller factors for both shells is ob-
tained, now dealing with just one type of S and one type of C neigh-
bour. The Mo–S distance is slightly longer compared to the
structure before reaction, consistent with the increases trans-influ-
ence of Me versus chloride, the former being the much stronger
donor.
r-
The Br3 analogue, however, shows a slightly different behav-
iour. Upon reacting 10 equivalents of AlMe3 with [MoBr3(SBz)],
no reaction at all is observed for reactions times up to 1 h. Increas-
ing the ratio Mo complex: AlMe3 to 1:20 (as with the complexes
above), significant changes in the EXAFS are observed, but much
smaller than for the tri-chloride systems. Comparing the EXAFS
of the complex before and after reaction shows a decrease in EXAFS
intensity only in the region between 5 < k < 11 Åꢂ1, which is ex-
actly the region where the Mo–Br is most significant (Mo–Br EXAFS
intensity peaks around k = 8, whereas the Mo–S contribution is
high at low k and diminishes towards higher k-values). Refining
the end-state EXAFS data lead to a good fit of two shells, one
Mo–S with a coordination number of 2.8(5) and a Mo–Br with a
coordination number of 0.8(1) (without a change in distances or
Debye Waller factors compared to the solution structure). Upon
reaction with AlMe3 the [MoBr3(SBz)] complex loses its halide li-
gand, but at a very different timescale than the chloride analogue,
i.e. the X/Me exchange kinetics are much slower going from the
chloride to the bromide ligand (Fig. 5). The mechanism, however,
seems comparable to the halide being abstracted by the AlMe3
activator. The overall coordination of Mo seems too low to be real-
istic, but no further shell could be introduced reliably (to improve
the fit significantly). Although there are small changes in EXAFS
visible at low k-values, which might indicate the introduction of
a light scatterer (C/N), the quality of the data does not allow us
to fit a reliable and significant Mo–C contribution here. It is possi-
ble that the Mo–C contribution is disordered (high Debye Waller
factor) and therefore too low in intensity to be determined under
the strong Mo–S and Mo–Br scatterers (with the quality of the data
we have). Similarly, the inclusion of a Mo–Mo contribution, in
search for either a direct bonded Mo dimer [35] or a halide-bridged
dimer, which has been suggested in related literature [12,13], was
unsuccessful and not significant based on the data as obtained, but
cannot be excluded completely. We thus most likely have a formed
a mononuclear mixed alkyl and halide species (Fig. 5), as was sug-
gested before for selected Cr systems under specific conditions
[13]. For all reactions, at longer timescales >1.5 h, the Mo K-edge
XAS signal disappears completely and black precipitate is observed
in the cell, indicating the formation of Mo metal (Mo–Mo contribu-
tions observed in time-resolved experiment, via infra).
3.2. ‘End-state’ analyses
The so-called end-states of the reactions, i.e. Mo(III) complex
activated with AlMe3, were characterised in the large solution cells.
The reactions were performed in this cell by injecting the required
amount of AlMe3 (solution in hexane or toluene) to the complex in
toluene solution. The EXAFS data acquisition for all these systems
starts about 5 min after injecting the AlMe3, with an overall acqui-
sition time of 35 min. The reactions were also followed with on-
the-fly XAS, acquisition of one spectrum per minute, and no further
changes were observed in this period of time (5–35 min) in this
cell. This indicates that a true reaction end-state has been reached
and measured. At longer timescales (>1.5 h), a clear precipitation of
Mo metal could be observed for all mixtures, indicating a second,
irreversible, reaction taking place. These reaction solutions were
carefully checked for X-ray beam damage, but no effects could be
observed. The solutions undergo the same colour changes without
exposure to the X-ray beam, and Mo precipitation is observed at
the same timescale (see also in ‘3.3 Time-Resolved XAS (and UV/
Vis)’).
The EXAFS analyses results of the reaction end points are again
represented in Tables 4–6. The [MoCl3(SNS-Decyl)] complex was
reacted with 20 equivalents of AlMe3. The EXAFS data can be re-
fined to two shells, i.e. a contribution of 3(1) N/C neighbours at
2.22(9) Å and a Mo–S shell of 1.9(9) sulphurs at 2.50(3) Å. As with
the solid state structure and solution, the Mo–N/C contribution is
difficult to determine and has the highest inaccuracy (large error
on coordination number and distance, with a large Debye Waller
factor). The Mo–S shell is well defined. However, between the solu-
tion before reaction and after (i.e. end-state), a clear decrease in
overall intensity in EXAFS can be observed which corresponds to
the loss in chlorine ligands. The SNS ligand remains intact, which
is confirmed by the presence of the Mo–S coordination at a dis-
tance similar to the solution structure, and the presence of a
Mo–N/C coordination. The increase in Mo–N/C coordination num-
ber can be attributed to the replacement of the chloride ligands by
methyl groups of the AlMe3. The coordination number is not very
well defined 3 1, but to achieve a full coordination sphere around
the Mo, a [Mo(CH3)3(SNS-Decyl)] complex is likely to be formed,
with one neighbour being the N atom of the SNS ligand and three
more neighbours corresponding to carbon from the AlMe3 moiety
(see Fig. 5). This ‘activated’ Mo structure is in agreement with anal-
ogous [CrCl3(L)] complexes upon activation with AlMe3 [14].
Activation of [MoCl3(SBz)] with 20 equivalents of AlMe3 (Table
5, Fig. 6) leads to a similar end-state-structure [Mo(CH3)3(SBz)]
with 3 sulphur atoms at 2.55(1) Å and 3 carbon atoms at 2.23(1)
3.3. Time-resolved XAS (and UV/Vis)
Time-resolved reactions were performed using the stopped-
flow instrument. The time-resolved XAS experiments were per-
formed in transmission, through the observation cuvette, with
time-resolved UV/Vis taken simultaneously, using optical fibres
in transmission perpendicular to the XAS observation. Because of
the different sensitivity of the two techniques, the observation
path-length of the XAS was 10 times the path-length of the UV–
Vis. The effect of the X-ray beam on the solution was carefully
checked by performing the reactions with and without X-ray beam
and observing the time-resolved UV/Vis spectra. No changes in the
time-resolved UV/Vis spectra were observed upon performing the
reactions in the X-ray beam, suggesting we are genuinely looking
at the chemical reaction taking place.

S.A. Bartlett et al. / Journal of Catalysis 284 (2011) 247–258
255
Fig. 5. Schematic representation of the proposed Mo complexes formed after activation with AlMe3 (E = S with R = CH2–C6H4-p-C(CH3)3 or E = N with R = n-C10H21).
the UV/Vis measurements with the EXAFS measurements is very
useful in establishing when changes in speciation occur and hence
guiding when EXAFS data should be recorded. After a reaction time
of 18 min in the stopped-flow, the Mo K-edge XAS signal as well as
the UV/Vis signal are lost completely, indicating the precipitation
of Mo. It is interesting to note that the precipitation of Mo metal
is observed at much shorter timescales compared to the dropout
observed in the large solution cell (18 min compared to 1.5 h). This
is probably due to the much smaller volume of the stopped-flow
cuvette compared to the large cell (30 lL and 3 mL, respectively).
The effects of X-ray beam were not significant as the same precip-
itation rates were observed when performing the reaction without
the X-ray beam (which was checked for both the large solution cell
and for the stopped-flow, as mentioned).
The reaction of [MoCl3(SBz)] was also performed in the
stopped-flow with different ratios of AlMe3 and followed with
QEXAFS (1 spectrum/min). When the reaction is performed at the
high MoCl3(SBz): AlMe3 ratio of 1:20 as was performed in the large
solution cell, the immediate precipitation of Mo was observed in
the stopped-flow cuvette, hindering the spectroscopy. Lower
amounts of AlMe3 were therefore investigated. At a stoichiometric
amount of MoCl3(SBz): AlMe3 of 1:1, no reaction was observed.
When the amount of AlMe3 was increased to 4 and 8 equivalents,
changes in the EXAFS upon reaction were observed, at the same
timescales. The EXAFS analysis of the 8 equivalent AlMe3 data after
4 min of reaction is presented in Table 5. A full displacement of the
chlorine ligand by carbon is observed, as for the end-state measure
in the large solution cell. However, at 4 min already a significant
proportion of Mo–Mo contributions can be detected (Fig. 6, Table
5), indicating a significant amount of Mo colloid formation.
Although the accuracy of the higher Mo–Mo shells is low, the Mo
particles detected should be of considerable size for these shells
to be visible and significant in the Fourier transforms. Again, this
irreversible deactivation of the Mo is observed at much shorter
timescales compared to the reactions in the large solution cell
(see Fig. 6). As said, the higher equivalents of AlMe3 (1:20 as in
the large solution cell, and 1:32) led to an almost immediate Mo
precipitation, indicating the reaction rate for this Mo complex, of
at least the deactivation part, is much faster in the stopped-flow
than in the large solution cell.
Fig. 6. Fourier transforms of k3-weighted EXAFS data on [MoCl3(SBz)] (in solution,
black line), with AlMe3 in large solution cell (20 equivalents, ꢀ40 min, blue line)
and stopped-flow (8 equivalents, ꢀ4 min, red line) (2 < k < 16 Åꢂ1). (For interpre-
tation of the references to colour in this figure legend, the reader is referred to the
web version of this article.)
The combined time-resolved data from the reaction of
[MoCl3(SNS)] (25 mM in toluene) with 20 equivalents of AlMe3
(identical to the reaction performed in the large solution cell) are
displayed in Fig. 7. QEXAFS spectra and time-resolved UV/Vis spec-
tra shown here were obtained every 20 s. Because of the slow pro-
gress of the reaction and to enhance the clarity of the data and its
changes, only a selection of spectra is presented here.
Although the overall EXAFS data quality is much lower (as ex-
pected due to the much shorter acquisition times), the overall
change in EXAFS in time is clear and identical to the end-state solu-
tion as discussed above, i.e. an overall loss in EXAFS intensity indi-
cating a loss of ligand. It is interesting to observe that most of the
changes are happening in the first 1.5 min, which is confirmed by
the UV/Vis data obtained simultaneously. The UV/Vis spectra of
the Mo starting complexes each show the lowest band around
4
4
22,000 cmꢂ1, attributed to the A2g to T2g transition of a Mo(III)
complex in approximately Oh symmetry. This corresponds to an in-
crease in Dq of approximately 50% relative to the corresponding
Cr(III) complexes [2,9,14,36], which is reasonable on the basis of
their respective 4d3 and 3d3 configurations. Based on Jorgensen’s
model [37], the lowest energy charge transfer transitions are
expected to occur >30,000 cmꢂ1. Upon addition of AlMe3 the d–d
transition decreases in intensity, with concomitant appearance of
new bands over a period of a few minutes (Fig. 7), and then disap-
pearance of all bands occurring around 18 min. The combination of
Reactions of the tri-bromide analogues were, as also observed in
the large solution cell, significantly slower than the tri-chloride
complexes. The [MoBr3(SBz)] complex was reacted with 20 equiv-
alents of AlMe3, followed with time-resolved XAS, and the EXAFS
after 4 min of reaction was analysed and presented in Table 6. As
for the end-state analyses, only a small loss in bromide ligand
was observed, whereas the SBz ligand remains intact and no signif-
icant Mo–C can be fitted reliably. The amount of Mo–Br contribu-
tions still present is slightly higher compared to the large solution

256
S.A. Bartlett et al. / Journal of Catalysis 284 (2011) 247–258
Fig. 7. Combined time-resolved UV–Vis and QEXAFS data on the reaction of [MoCl3(SNS)] (25 mM in toluene) with 20 equivalents of AlMe3 at t = 0 (black, t = 1 min (red),
t = 2.5 min (blue), t = 7 min (green), t = 18 min (orange), t = 19 min (pink). (For interpretation of the references to colour in this figure legend, the reader is referred to the web
version of this article.)
cell, which corresponds to the shorter reaction time in the stopped-
flow. Here, no clear effect upon reaction rate is observed going
from the large volume large solution cell to the small volume
stopped-flow cuvette.
logue allows us to distinguish between ligand and halide easier
(compared to chlorine complex). As for the time-resolved data, a
small decrease in the EXAFS, especially in the 5 < k < 11 Åꢂ1 region,
in time is observed. Comparing the Fourier transforms of the three
data sets shows nicely the decrease in intensity on the high R side,
i.e. where the Mo–Br contribution is. The analyses of both data sets,
as presented in Table 6, show again the partial removal of Br, with-
out dissociation of the SBz ligand, which closely corresponds to the
time-resolved data. In addition, the freeze-quench set-up now
gives access to good quality EXAFS data 5 s after mixing, something
which could not be obtained using QEXAFS as performed (Mo K-
edge QEXAFS time resolution possible at our beamtime periods
were 20–30 s/spectrum, or 1–3 min/spectrum).
The combination of experiments as performed in this study
clearly show the stepwise displacement of the halide ligand upon
activating [MoX3(L)] complexes with AlMe3, forming mixed alkyl
and halide mononuclear complexes, see Fig. 5, as was suggested
for some Cr systems [13]. The high Mo K-edge energy allows
time-resolved stopped-flow XAS experimentation down a time res-
olution of 20 s per spectrum (for good S/N quality data to allow EX-
AFS analyses). The new freeze-quench methodology is an
extension to the stopped-flow with a freeze-quenching time of less
than a second, thereby giving access to a time regime not accessi-
ble with the time-resolved XAS as performed.
3.4. Freeze-quench study of the alkylation of [MoBr3(SBz)]
The BioLogic stopped-flow quench system was modified and an
air and moisture tight freeze-quench accessory developed to allow
us to use the millisecond time resolution of the stopped-flow appa-
ratus to select and isolate different intermediates, freeze-quench
them and keep them frozen to allow long EXAFS experimentation
to be conducted (if required). The reaction solutions are sprayed di-
rectly into a KaptonÒ tube immersed in liquid nitrogen. The re-
quired freezing time for the volume of 75 lL is very difficult to
assess and measure, but estimated to be less than 1 s in total.
The entire system was thoroughly tested with a very air and mois-
ture sensitive system [(TiCp2)2(l-Cl)2] (green) [38], to ensure its
functionality. The deep green dimeric Ti solution (in toluene) sys-
tem is very sensitive to O2 and rapidly changes to yellow upon oxi-
dation. Glove box tests showed that the solid oxidised at O2 levels
>5 ppm. After taking the solution through the stopped-flow and
freezing it into the KaptonÒ tube, the solution remained green,
indicating a O2 free system was established.
Although the Mo complexes have shown a very low catalytic
activity towards ethene oligomerisation [22], it is reasonable that
the Mo systems undergo similar chemistry in the early stages of
activation. This is in fact what we see. The stepwise displacement
of the halide forming [Mo(CH3)3(L)] is in agreement with our pre-
vious studies on the AlMe3 activation of comparable [CrCl3(L)]
complexes [14]. Whereas literature reports on halide-bridged di-
mer formation at low Cr:Al ratios [12], dimeric Mo species are
not observed in this study at high Mo:Al ratios, i.e. ratios more rel-
evant to industrial conditions [2]. Going from the Cr to the Mo ana-
logues, a clear decrease in reaction rate was observed, as expected.
However, also a significant dependence of reaction kinetics is ob-
served upon halide ligand, with the [MoCl3(L)] being much faster
than the [MoBr3(L)] analogues. It is difficult to immediately ratio-
nalise this difference, since the exact substitution mechanism is
unknown and may in fact be different for the two types of com-
plexes. The reaction can be based upon a dissociative or Lewis acid
approach, with Al–Cl versus Al–Br bond formations being the rate
determining steps. Moreover, at the high concentrations of reac-
tants as used in this study, the activation reaction is in competition
with the complex dissociation reaction as observed, especially for
To validate the freeze-quench approach, we have investigated
the reaction of [MoBr3(SBz)] with 20 equivalents of AlMe3. Ini-
tially, the starting Mo solution is injected in the capillary, frozen
and measured. Good quality transmission and fluorescence data
could be obtained for these Mo systems, with identical EXAFS spec-
tra. Analysis of the EXAFS revealed that the same structural param-
eters were obtained as for the solution state structure. The reaction
was then freeze-quenched after 5 s and again after 5 min of reac-
tion, to compare to the time-resolved data as obtained in the
stopped-flow. Fig. 8 shows the EXAFS data obtained for the differ-
ent frozen intermediates alongside the solution starting structure.
The signal to noise of the frozen data is somewhat lower compared
to the time-resolved data in the cuvette. This can be easily im-
proved by acquiring more spectra and measuring longer (but this
was not done here due to time constraints). In practice, it turns
out (above) that this reaction is rather slow, so quick freezing did
not necessarily reveal any new intermediates. It did however allow
us to probe the reaction at identical conditions (20 equivalents
AlMe3) as the time-resolved XAS. We should then in principle be
observing the same intermediates and thus prove the functionality
of the stopped-flow freeze-quench. Moreover, the bromide ana-

S.A. Bartlett et al. / Journal of Catalysis 284 (2011) 247–258
257
Fig. 8. Raw k2-weighted EXAFS data (a) and Fourier Transform (1.8 < k < 12) (b) data on the [MoBr3(SBz)] solution (black), with 20 equivalents of AlMe3 FQ after 5 s (red) and
FQ after 5 min (blue) samples. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
the chloride analogues. This is not necessarily a problem for the
catalysis, since we are only looking at the first stage of activation
here, and all subsequent catalysis reaction steps with ethene may
make the formed complexes more stable. This study does provide
insights in the low overall catalytic activity and deactivation of
these complexes, as observed in catalytic tests.
can be prevented or at least decreased by keeping the samples at
low temperature. Further experiments using this approach to
examine the catalytic process associated with the [CrCl3(SNS)] eth-
ene timerisation system are now planned.
Acknowledgments
The ESRF is gratefully acknowledged for provision of beamtime
(26-01-829) and the staff with their help. The Swiss Light Source,
PSI, Villigen, Switzerland is gratefully acknowledged for provision
of beamtime, with EU funding under EU Project FP7 ELISA. The Dia-
mond Light Source for beamtime provision under project SP7046-
1. MT gratefully acknowledges the EPSRC (EP/E060404/1), SAB and
PPW the EPSRC (EP/F032463/1). Dr. M. Webster is gratefully
acknowledged for help with the X-ray crystallography.
4. Conclusions and outlook
We have used time-resolved Mo K-edge EXAFS in a stopped-
flow system to probe the first stage of activation of the Mo ana-
logues of the industrially important ethene trimerisation catalysts
and compared the results with start and end-state studies. The re-
sults demonstrate that treatment of the Mo complex solution with
excess AlMe3 leads to halide loss, forming [Mo(CH3)3(L)], and pro-
vides strong evidence that dinuclear species with direct Mo–Mo
bonds or halide-bridged dimers are not present under the condi-
tions examined (1:20 Mo:Al). At the high AlMe3 ratios used, with
the complexes investigated, the activation reaction is in competi-
tion with the deactivation reaction, especially for the trichloride
complexes, providing insights in the low overall catalytic activity
and deactivation as has been reported for these systems under real
catalytic conditions [21]. Whereas the use of the stopped-flow for
time-resolved homogeneous catalysis has been reported before
[17,18,20], we now have also developed a freeze-quench attach-
ment which takes advantage of the excellent volume and timescale
control offered by the stopped-flow instrument, and allows reac-
tion mixtures to be frozen (to liquid nitrogen temperature) within
1 s of mixing under anaerobic conditions. This allows reactive,
short-lived intermediate species to be trapped and enables EXAFS
data collections to be conducted on these over long timescales
(as required for 3d transition metals with low X-ray absorption
edge energies and for low concentrations). The freeze-quench
experiment can be performed in fluorescence mode, which is
essential for studying systems with low energy absorption edges
(e.g. Cr, Sc, etc.). A key advantage is that the much longer data
acquisition times associated with these 3d metal systems are easily
accommodated. This technique has great potential for probing
reactive intermediates in many homogeneous catalytic systems,
down to the seconds timescale with the freeze-quench procedure
taking about 1 s. In fact, the methodology increases the XAS time
resolution for these systems with a factor of 4,000–10,000 (1 s
compared to 1–4 h as before). Not only reactions can be quenched
and measured, also in case of samples which are easily affected by
beam damage (which is not the case in this study), beam damage
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.jcat.2011.10.015.
References
[1] J. Skipinska, Chem. Rev. 91 (1991) 613.
[2] D.S. McGuinness, Chem. Rev. 111 (2011) 2321 (and references therein).
[3] J.T. Dixon, M.J. Green, F.M. Hess, D.H. Morgan, J. Organometal. Chem. 689
(2004) 3641.
[4] D.F. Wass, Dalton Trans. (2007) 816.
[5] W. Janse van Rensburg, C. Grove, J.P. Steynberg, K.B. Stark, J.J. Huyser, P.J.
Steynberg, Organometallics 23 (2004) 1207.
[6] (a) D.S. McGuinness, P. Wasserscheid, W. Keim, J.T. Dixon, J.J.C. Grove, C. Hu, U.
Englert, Chem. Commun. (2003) 334;
(b) D.S. McGuinness, P. Wasserscheid, W. Keim, D.H. Morgan, J.T. Dixon, A.
Bollmann, H. Maumela, F.M. Hess, U. Englert, J. Am. Chem. Soc. 125 (2003)
5272;
(c) D.S. McGuinness, P. Wasserscheid, D.H. Morgan, J.T. Dixon, Organometallics
24 (2005) 552.
[7] A. Carter, S.A. Cohen, N.A. Cooley, A. Murphy, J. Scutt, D.F. Wass, Chem.
Commun. (2002) 858.
[8] E.Y.-X. Chen, T.J. Marks, Chem. Rev. 100 (2000) 1391.
[9] D.S. McGuinness, D.B. Brown, R.P. Tooze, F.M. Hess, J.T. Dixon, A.M.Z. Slawin,
Organometallics 25 (2006) 3605.
[10] D.S. McGuinness, A.J. Rucklidge, R.P. Tooze, A.M.Z. Slawin, Organometallics 26
(2007) 2561.
[11] A.J. Rucklidge, D.S. McGuinness, R.P. Tooze, A.M.Z. Slawin, D.A. Pelletier, M.J.
Hanton, P.B. Webb, Organometallics 26 (2007) 2782.
[12] (a) C. Temple, A. Jabri, P. Crewdson, S. Gambarotta, I. Korobkov, R. Duchateau,
Angew. Chem. Int. Ed. 45 (2006) 7050;
(b) A. Jabri, C. Temple, P. Credson, S. Gambarotta, I. Korobkov, R. Duchateau, J.
Am. Chem. Soc. 128 (2006) 9238.
[13] S. Lui, R. Pattacini, P. Braunstein, Organometallics 30 (2011) 3549.

258
S.A. Bartlett et al. / Journal of Catalysis 284 (2011) 247–258
[14] J.O. Moulin, J. Evans, D.S. McGuinness, G. Reid, A.J. Rucklidge, R.P. Tooze, M.
Tromp, Dalton Trans. (2008) 1177.
(b) G.M. Sheldrick, SHELXL-97, Program for Crystal Structure Refinement,
University of Göttingen, Germany, 1997;
[15] N. Yoshida, T. Matsushita, S. Saigo, H. Oyanagi, H. Hashimoto, M. Fujimoto, J.
Chem. Soc. Chem. Commun. (1990) 354.
[16] M.B.B. Abdul Rahman, P.R. Bolton, J. Evans, A.J. Dent, I. Harvey, S. Diaz-Moreno,
Faraday Discuss. 122 (2002) 211.
(c) Flack, Acta Crystallogr. Sect. A 39 (1983) 876.
[29] M. Tromp, ‘Developments of Time-Resolved XAFS Techniques: Applications in
Homogeneous Catalysis’, Ph.D. Thesis, Utrecht, The Netherlands, 2004, pp. 20–
23.
[17] M. Tromp, J.R.A. Sietsma, J.A. van Bokhoven, G.P.F. van Strijdonck, R.J. van
Haaren, A.M.J. van der Eerden, P.W.N.M. van Leeuwen, D.C. Koningsberger,
Chem. Commun. (2003) 128.
[30] S. Nikitenko, A.M. Beale, A.M.J. van der Eerden, S.D.M. Jacques, O. Leynaud,
M.G. O’Brien, D. Detollenaere, R. Kaptein, B.M. Weckhuysen, W. Bras, J.
Synchrotron Radiat. 15 (2008) 632–640.
[18] M. Tromp, S.S. van Berkel, A. van den Hoogenband, M.C. Feiters, B. de Bruin,
S.G. Fiddy, J.A. van Bokhoven, P.W.N.M. van Leeuwen, G.P.F. van Strijdonck,
D.C. Koningsberger, Organometallics 29 (2010) 3085.
[19] J. Evans, A. Puig-Molina, M. Tromp, MRS Bull. 32 (2007) 1038.
[20] G. Smolentsev, G. Guilera, M. Tromp, S. Pascarelli, A.V. Soldatov, J. Chem. Phys.
130 (2009) 174508.
[31] A.J. Dent, G. Cibin, S. Ramos, A.D. Smith, S.M. Scott, L. Varandas, M.R. Pearson,
N.A. Krumpa, C.P. Jones, P.E. Robbins, J. Phys.: Conf. Ser. 190 (2009) 01203.
[32] M.J. Newville, J. Synchrotron Radiat. 8 (2001) 322.
[33] B. Ravel, M.J. Newville, J. Synchrotron Radiat. 12 (2005) 537.
[34] (a) C.D. Beard, L. Carr, M.F. Davis, J. Evans, W. Levason, L.D. Norman, G. Reid, M.
Webster, Eur. J. Inorg. Chem. (2006) 4399;
[21] S.P. Downing, M.J. Hanton, A.M.Z. Slawin, R.P. Tooze, Organometallics 28
(2009) 2417.
(b) G.J. Grant, K.E. Rogers, W.N. Setzer, D.G. VanDerveer, Inorg. Chim. Acta 234
(1995) 35;
[22] O. Kleifeld, A. Frenkel, J.M. Martin, I. Sagi, Nat. Struct. Biol. 10 (2003) 98.
[23] F. Stoffelbach, D. Saurenz, R. Poli, Eur. J. Inorg. Chem. (2001) 2699.
[24] B.E. Owens, R. Poli, A.L. Rheingold, Inorg. Chem. 28 (1989) 1456.
[25] D.S. McGuinness, P. Wasserscheid, W. Kiem, D.H. Morgan, J.T. Dixon, A.
Bollmann, H. Maumela, F.M. Hess, U. Englert, J. Am. Chem. Soc. 125 (2003)
5272.S3.
(c) R.E. Wolf, J.R. Hartman, J.M.E. Storey, B.M. Foxman, S.R. Cooper, J. Am.
Chem. Soc. 109 (1987) 4328.
[35] F.A. Cotton, Multiple bonds between metal atoms, in: F.A. Cotton, C.A. Murillo,
R.A. Walton (Eds.), third ed., Springer, 2005. ISBN 978-0-387-25084-7.
[36] J.O. Moulin, ‘X-Ray Absorption Spectroscopy for the Study of a Homogeneous
Catalysis System based Upon Chromium, Ph.D. Thesis, Southampton, United
Kingdom, 2006.
[26] E. Jordon, Anal. Chem. 40 (1968) 2150.
[27] D. Sellmann, L. Zapf, J. Organomet. Chem. 289 (1985) 57.
[28] (a) G.M. Sheldrick, SHELXS-97, Program for Crystal Structure Solution,
University of Göttingen, Germany, 1997;
[37] A.B.P. Lever, Inorganic Electronic Spectroscopy, second ed., Elsevier, Canada,
1984 (chapter 5).
[38] S.J. Nieter Burgmayer, J. Chem. Ed. 75 (1998) 460.