A Dendritic Effect in homogeneous catalysis with carbosilane-supported arylnickel(II) catalysts: Observation of active-site proximity effects in atom-transfer radical addition

Article abstract of DOI:10.1021/ja0026612

Transmetalation of polylithiated, carbosilane (CS) dendrimers functionalized with the potentially terdentate ligand [C₆H₂(CH₂ NMe₂)₂-2,6-R-4]^- ( = NCN) with NiCl₂(PEt₃)₂ produced a series of nickel-containing dendrimers [G0]-Ni₄ (4), [G1]-Ni₁₂ (5), and [G2]-Ni₃₆ (7) in moderate to good yields. The metallodendrimers 4, 5, and 7 are catalytically active in the atom-transfer radical addition (ATRA) reaction (Kharasch addition reaction), viz. the 1:1 addition of CCl₄ to methyl methacrylate (MMA). The catalytic data were compared to those obtained for the respective mononuclear compound [NiCl(C₆H₂{CH₂ NMe₂}₂-2,6-SiMe₃-4)] (2). This comparison indicates a fast deactivation for the dendrimer catalysts beyond generation [G0]. The deactivation of [G1]-Ni₁₂ (5) and [G2]-Ni₃₆ (7) is caused by irreversible formation of catalytically inactive Ni(III) sites on the periphery of these dendrimers. This hypothesis is supported by results of model studies as well as ESR spectroscopic investigations. Interestingly, the use of two alternative nickelated [G1] dendrimers [G1]^*-Ni₁₂ (11) and [G1]-Ni₈ (15), respectively, in which the distance between the Ni sites is increased, leads to significantly improved catalytic efficiencies which approximate those of the parent derivative 2 and [G0]-Ni₄ (4). Preliminary membrane catalysis experiments with [G0]-Ni₄ (4) and [G1]-Ni₁₂ (5) show that 5 can be efficiently retained in a membrane reactor system. The X-ray crystal structure of the Ni(III) complex [NiCl₂(C₆H₂ {CH₂NMe₂}₂-2,6-SiMe₃-4)] (16), obtained from the reaction of 2 with CCl₄, is also reported.

Full text of DOI:10.1021/ja0026612

12112
J. Am. Chem. Soc. 2000, 122, 12112-12124
A “Dendritic Effect” in Homogeneous Catalysis with
Carbosilane-Supported Arylnickel(II) Catalysts: Observation of
Active-Site Proximity Effects in Atom-Transfer Radical Addition
Arjan W. Kleij,† Robert A. Gossage,†, Robertus J. M. Klein Gebbink,† Nils Brinkmann,‡
Ed J. Reijerse,| Udo Kragl,‡ Martin Lutz,§ Anthony L. Spek,§,# and Gerard van Koten*,†
Contribution from the Debye Institute, Department of Metal-Mediated Synthesis, Utrecht UniVersity,
Padualaan 8, 3584 CH, Utrecht, The Netherlands, Forschungszentrum Ju¨lich GmbH,
Institut fu¨r Biotechnologie, D-52425 Ju¨lich, Germany, BijVoet Center for Biomolecular Research,
Department of Crystal and Structural Chemistry, Utrecht UniVersity, Padualaan 8, 3584 CH Utrecht,
The Netherlands, Department of Molecular Spectroscopy, UniVersity of Nijmegen, ToernooiVeld,
6525 ED, Nijmegen, The Netherlands
ReceiVed July 19, 2000
Abstract: Transmetalation of polylithiated, carbosilane (CS) dendrimers functionalized with the potentially
terdentate ligand [C6H2(CH2NMe2)2-2,6-R-4]- ( ) NCN) with NiCl2(PEt3)2 produced a series of nickel-
containing dendrimers [G0]-Ni4 (4), [G1]-Ni12 (5), and [G2]-Ni36 (7) in moderate to good yields. The
metallodendrimers 4, 5, and 7 are catalytically active in the atom-transfer radical addition (ATRA) reaction
(Kharasch addition reaction), viz. the 1:1 addition of CCl4 to methyl methacrylate (MMA). The catalytic data
were compared to those obtained for the respective mononuclear compound [NiCl(C6H2{CH2NMe2}2-2,6-
SiMe3-4)] (2). This comparison indicates a fast deactivation for the dendrimer catalysts beyond generation
[G0]. The deactivation of [G1]-Ni12 (5) and [G2]-Ni36 (7) is caused by irreversible formation of catalytically
inactive Ni(III) sites on the periphery of these dendrimers. This hypothesis is supported by results of model
studies as well as ESR spectroscopic investigations. Interestingly, the use of two alternative nickelated [G1]
dendrimers [G1]*-Ni12 (11) and [G1]-Ni8 (15), respectively, in which the distance between the Ni sites is
increased, leads to significantly improved catalytic efficiencies which approximate those of the parent derivative
2 and [G0]-Ni4 (4). Preliminary membrane catalysis experiments with [G0]-Ni4 (4) and [G1]-Ni12 (5) show
that 5 can be efficiently retained in a membrane reactor system. The X-ray crystal structure of the Ni(III)
complex [NiCl2(C6H2{CH2NMe2}2-2,6-SiMe3-4)] (16), obtained from the reaction of 2 with CCl4, is also
reported.
Introduction
create new materials that bridge the gap between homog-
eneous and heterogeneous catalysis. In this regard, we pioneered
the first “metallodendrimer catalyst”, in which catalytically
active metal centers were connected to an inert carbosilane
framework via an appropriate (carbamate) linker group (see
Figure 1).6 These compounds were designed to take advantage
of the activity, selectivity, and ease of characterization and
modification of homogeneous catalysts in combination with the
size properties that are necessary for effective recovery (i.e.,
nanofiltration) of the catalytic species (cf., heterogeneous
systems).3b,7
Current research in our group concentrates on the use of
(carbosilane) dendrimers1-4 and hyperbranched polymers5 to
* To whom correspondence should be adressed. Telephone: +3130
2533120. Fax: +3130 2523615. E-mail: g.vankoten@chem.uu.nl.
† Department of Metal-Mediated Synthesis, Utrecht University.
‡ Institut fu¨r Biotechnologie.
§ Department of Crystal and Structural Chemistry, Utrecht University.
| University of Nijmegen.
Current address: Department of Chemistry, Acadia University, Wolfville,
Novia Scotia, Canada BOP1XO.
# Address correspondence pertaining to crystallographic studies to this
author. E-mail: a.l.spek@chem.uu.nl.
(3) (a) Majoral, J. P.; Caminade, A. M. Chem. ReV. 1999, 99, 845-880
and references therein. (b) de Groot, D.; Eggeling, E. B.; de Wilde, J. C.;
Kooijman, H.; Haaren, R. J.; van der Made, A. W.; Spek, A. L.; Vogt, D.;
Reek, J. N. H.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. Chem. Commun.
1999, 1623. (c) Petrucci-Samija, Guillemette, V.; Dasgupta, M.; Kakkar,
A. K. J. Am. Chem. Soc. 1999, 121, 1968. (d) Alper, H.; Bourque, S. C.;
Manzer, L. E.; Arya, P. J. Am. Chem. Soc. 2000, 122, 956.
(4) (a) van der Made, A. W.; van Leeuwen, P. W. N. M. J. Chem. Soc.,
Chem. Commun. 1992, 1400-1401. (b) Zhou, L. L.; Roovers, J. Macro-
molecules 1993, 26, 963-968. (c) Roovers, J.; Zhou, L. L.; Toporowski,
P. M.; van der Zwan, M.; Iatrou, H.; Hadjichristidis, N. ibid. 1993, 26,
4324-4329. (d) van der Made, A. W.; van Leeuwen, P. W. N. M.; de Wilde,
J. C.; Brandes, R. A. C. AdV. Mater. 1993, 5, 466-468.
(1) For reviews on this subject, see: (a) Newkome, G.; Moorefield, C.
N.; Vo¨gtle, F. Dendritic Molecules - Concepts, Syntheses, PerspectiVes;
VCH: Weinheim, 1996. (b) Newkome, G., Ed. AdVances in Dendritic
Macromolecules; JAI Press: Greenwich, CT, 1995. For more recent reviews
in dendrimer chemistry, see: (c) Bosman, A. W.; Janssen, H. M.; Meijer,
E. W. Chem. ReV. 1999, 99, 1665. (d) Newkome, G.; He, E.; Moorefield
Chem. ReV. 1999, 99, 1689. (e) Fischer, M.; Vo¨gtle, F. Angew. Chem., Int.
Ed. 1999, 38, 885 (f) Frey, H.; Lach, C.; Lorenz, K. AdV. Mater. 1998, 10,
279. (g) Gorman, C. ibid. 1998, 10, 295. (h) Gudat, D. Angew. Chem., Int.
Ed. Engl. 1997, 36, 1951. (i) Fre´chet, J. M. J.; Hawker, C. J.; Gitsov, I.;
Leon, J. W. J. Macromol. Sci. Pure Appl. Chem. 1996, A33, 1399. Dendritic
compounds have been made primarily from carbon-based materials, although
the “branching points” are frequently heteroatoms such as nitrogen2,
phosphorus,3 and silicon.4
(5) Schlenk, C.; Kleij, A. W.; Frey, H.; van Koten, G. Angew. Chem.,
Int. Ed. 2000, 39, 3445-3447.
(6) Knapen, J. W. J.; Van der Made, A. W.; de Wilde, J. C.; van Leeuwen,
P. W. N. M.; Wijkens, P.; Grove, D. M.; van Koten, G. Nature 1994, 372,
659-663.
(2) (a) Tomalia, D. A.; Taylor, A. M.; Goddard, W. A., III. Angew. Chem.
1990, 102, 119-156, Angew. Chem., Int. Ed. Engl. 1990, 29, 138-175.
(b) Tomalia, D. A. Sci. Am. 1995, 272, 62-70.
10.1021/ja0026612 CCC: $19.00 © 2000 American Chemical Society
Published on Web 11/22/2000

Carbosilane-Supported Arylnickel(II) Catalysts
J. Am. Chem. Soc., Vol. 122, No. 49, 2000 12113
of the sterical crowding at the periphery of the CS-dendritic
species and, in particular, the mutual distance between the Ni
catalytic sites. We have further explored the catalytic activity
of these new metallodendrimers in the atom-transfer radical
addition reaction and compared the turnover numbers and
electrochemical properties of these materials with the respective
mononuclear analogue [NiCl(C6H2{CH2NMe2}2-2,6-R-4)] (Fig-
ure 2, R ) SiMe3).
Figure 1. Schematic representation of a metallodendritic catalyst.6
In an earlier preliminary communication, we have reported
that the higher-generation metallodendritic catalysts have dra-
matically lower catalytic activities and that the Ni sites no longer
act as independent catalytic units.12 However, it was also shown
that an increase in spatial separation between the nickel sites
brought about by the use of sterically less congested CS-
dendritic supports, led to retention of catalytic activity. The
underlying mechanistic reasons for the deactivation of the
higher-generation nickel-containing dendrimers have been stud-
ied in more detail and are described herein. Moreover, recent
results demonstrate that modern ultrafiltration membrane reac-
tors can be used to effectively recover the dendrimer catalysts
and hence can be employed for continuous-operation applica-
tions in homogeneous catalysis.13
Figure 2. Schematic structure of para-functionalized nickel complexes
derived from the NCN ligand.
The original dendrimer catalysts gave comparable activity
per metal atom when compared with the monomeric analogues
(Figure 2, X ) Br).8 Specifically, these Ni-based complexes
are active catalysts for the atom-transfer radical (i.e., Kharasch)
addition of polyhalogenated alkanes to olefins9 and are char-
acterized by the incorporation of a bis(ortho) chelating monoan-
ionic “pincer” ligand. This fragment contains a formal aryl
carbanion in combination with two trans-positioned tertiary
amine donor groups and is derived from the simple para-
substituted arenes 1-R-3,5-bis[(dimethylamino)methyl]benzene
(Figure 2).8 Ligands of this class are currently an area of active
study due to the unique catalytic properties of a variety of metal
“pincer” complexes.10
Results and Discussion
Preparation of Nickelated Carbosilane Dendrimers. Re-
cently, we reported the synthesis of carbosilane dendrimers
functionalized with terminal bidentate (C,N) or tridentate
(N,C,N) ligand fragments.11 These systems could be selectively
and quantitatively converted into their corresponding polylithi-
ated analogues. We anticipated that these lithiated carbosilanes
would be useful starting materials for the synthesis of metal-
lodendrimers containing various (N)CN-ligated transition metal
fragments. Therefore, we extended these earlier studies and
prepared a series of novel nickel-containing carbosilane den-
drimers with the NCN-NiCl fragments attached to the dendritic
backbone via a direct, stable Si-C bond. We prepared the model
compounds [NiCl(C6H2{CH2NMe2}2-2,6-SiMe3-4)] (2) and
Me2Si[C6H2{CH2NMe2}2-3,5-(NiCl)-4]2 (3) to serve as refer-
ence compounds in catalysis and for electrochemical investiga-
tions using cyclic voltammetry.
The mono-nickelated derivative [NiCl(C6H2{CH2NMe2}2-2,6-
SiMe3-4)] (2) was obtained by a two-step reaction sequence.
Lithiation of 1-trimethylsilyl-3,5-bis[(dimethylamino)methyl]-
benzene14 with t-BuLi in hexane at room temperature (RT) for
18 h, followed by treatment of the lithiated intermediate with
[NiCl2(PEt3)2]15 affords 2 as an orange solid (61% yield). The
bis-nickelated derivative Me2Si[C6H2{CH2NMe2}2-3,5-(NiCl)-
4]2 (3) was produced in a 66% yield by a similar approach using
the silane derivative Me2Si[C6H3{CH2NMe2}2-3,5]2 (1) as
starting material. This latter compound was obtained as a yellow
oil and isolated in quantitative yield by treatment of Me2SiCl2
with an excess of the lithium reagent [Li(C6H3{CH2NMe2}2-
3,5)]11a ( ) Li-NCN, Scheme 1). The dendrimer analogues of
In this report, we detail a novel synthetic methodology for
the production of metallodendrimer catalysts via a simple
procedure involving attachment of the “pincer” ligand to the
dendrimer periphery via a Si-C bond. The resulting pincer-
containing, dendrimer ligands were selectively lithiated and
transmetalated,11 affording catalytic units which are bound to
the dendritic framework via a direct Si-C bond. This latter
feature makes the metallodendrimers chemically even more
robust than our earlier systems (Figure 1). However, removal
of the carbamate-based linker will dramatically affect the nature
(7) (a) Hovestad, N. J.; Eggeling, E. B.; Heidbu¨chel, H. J.; Jastrzebski,
J. T. B. H.; Kragl, U.; Keim, W.; Vogt, D.; van Koten, G. Angew. Chem.,
Int. Ed. 1999, 111, 1763. (b) Brinkmann, N.; Giebel, D.; Lohmer, G.; Reetz,
M. T.; Kragl, U. J. Catal. 1999, 183, 163-168. (c) Giffels, G.; Beliczey,
J.; Felder, M.; Kragl, U. Tetrahedron: Assymmetry 1998, 9, 691-696.
(8) (a) van de Kuil, L. A.; Luitjes, H.; Grove, D. M.; Zwikker, J. W.;
van de Linden, J. G. M.; Roelofsen, A. M.; Jenneskens, L. W.; Drenth,
W.; van Koten, G. Organometallics 1994, 13, 468-477. (b) Grove, D. M.;
Verschuuren, A. H. M.; van Koten, G. J. Mol. Catal. 1988, 45, 169. (c)
Grove, D. M.; Verschuuren, A. H. M.; van Koten, G.; van Beek, J. A. M.
J. Organomet. Chem. 1989, 372, C1-C6.
(9) (a) Gossage, R. A.; van de Kuil, L. A.; van Koten, G. Acc. Chem.
Res. 1998, 31, 423. (b) van de Kuil, L. A.; Grove, D. M.; Gossage, R. A.;
Zwikker, J. W.; Jenneskens, L. W.; Drenth, W.; van Koten, G. ibid. 1997,
16, 4985-4994. (c) Kharasch, M. S.; Jensen, E. V.; Urry, W. H. Science
1945, 102, 128.
(10) For NCN-type ligands, see: (a) van Koten, G. Pure Appl. Chem.
1989, 61, 1681-1694. (b) Rietveld, M. H. P.; Grove, D. M.; van Koten,
G. New J. Chem. 1997, 21, 751-771. (c) Stark, M. A.; Jones, G.; Richards,
C. J. Organometallics 2000, 19, 1282. For PCP-type ligands, see: (d) Dani,
P.; Karlen, T.; Gossage, R. A.; Gladiali, S.; van Koten, G. Angew. Chem.,
Int. Ed. 2000, 39, 743. (e) Gupta, M.; Hagen, C.; Kaska, W.; Cramer, R.
E.; Jensen, C. M. J. Am. Chem. Soc. 1997, 119, 840. (f) Lee, D. W.; Kaska,
W. C.; Jensen, C. M. Organometallics 1998, 17, 1. (g) Jensen, C. M. Chem.
Commun. 1999, 2443. (h) Liu, F.; Goldman, A. S. Chem. Commun. 1999,
655. (i) Longmire, J. M.; Zhang, X. Organometallics 1998, 17, 4374. (j)
Gorla, F.; Togni, A.; Venanzi, L. M.; Albinati, A.; Lianza, F. Organome-
tallics 1994, 13, 1607. For SCS-type ligands, see: (k) Loeb, S. J.; Shimizu,
G. K. H. J. Chem. Soc., Chem. Commun. 1993, 1395. (l) Moulton, C. J.;
Shaw, B. L. J. Chem. Soc., Dalton Trans. 1975, 1020. (m) Bergbreiter, D.
E.; Osburn, P. L.; Liu, Y.-S. J. Am. Chem. Soc. 1999, 121, 9531. For a
review on C-C bond activation by “pincer” type complexes, see: (n)
Rybtchinski, B.; Milstein, D. Angew. Chem., Int. Ed. 1999, 38, 870.
(11) (a) Kleij, A. W.; Klein, H.; Jastrzebski, J. T. B. H.; Smeets, W. J.
J.; Spek, A. L.; van Koten, G. Organometallics 1999, 18, 268. (b) Kleij, A.
W.; Klein, H.; Jastrzebski, J. T. B. H.; Spek, A. L.; van Koten, G.
Organometallics 1999, 18, 277.
(12) For a preliminary communication of this work, see: Kleij, A. W.;
Gossage, R. A.; Jastrzebski, J. T. B. H.; Boersma, J.; van Koten, G. Angew.
Chem., Int. Ed. 2000, 39, 179.
(13) (a) Kragl, U. and Dreisbach, C. Angew. Chem., Int. Ed. Engl. 1996,
35, 642. (b) Kragl, U.; Dreisbach, C.; Wandrey, C. Membrane reactors in
Homogeneous Catalysis. Applied Homogeneous Catalysis with Organome-
tallic Compounds; Cornils, B., Herrmann, W. A., Eds.; VCH: Weinheim,
1996; pp 833-843.
(14) Steenwinkel, P.; James, S. L.; Grove, D. M.; Veldman, N.; Spek,
A. L.; van Koten, G. Chem. Eur. J. 1996, 2, 1440.
(15) This transmetallating reagent is easily prepared from the reaction
between NiCl2‚6H2O and PEt3 in EtOH. See: Jensen, K. A. Z. Anorg. Allg.
Chem. 1936, Band 229, 273.

12114 J. Am. Chem. Soc., Vol. 122, No. 49, 2000
Scheme 1: Synthesis of Model Compounds 2 and 3a
Kleij et al.
should be noted that dendrimer [G2]-Ni36 (7) is less soluble
than [G0]-Ni4 (4) and [G1]-Ni12 (5), indicating that beyond the
first generation the solubility behavior is increasingly controlled
by the peripheral organometallic groups (i.e., size). The
decreased solubility features found for 7 could explain its low
isolated yield.
In addition, we also prepared two less “condensed” [G1]
nickelated carbosilane dendrimers. These alternative [G1] den-
drimers were constructed to be able to vary the proximity
between the surface-bonded nickel sites and their accessibility
for substrate molecules (vide infra).
Two different approaches were used to increase the separation
between the external organonickel(II) chloride groups. The first
approach involves the introduction of an inert carbosilane tether
(i.e., -CH2CH2CH2Si(Me)2-) on the [G1] periphery. The syn-
thetic pathway leading to the extended nickel-containing den-
drimer 11 (abbreviated as [G1]*-Ni12) is outlined in Scheme 4.
The synthesis of [G1]*-Ni12 (11) begins with an allylation
reaction of the known chlorosilane [G1]-SiMe2Cl4a with an
excess of allylMgBr to yield the dodeca-olefin [G1]-SiMe2-
CH2CHdCH2 (8) as a viscous oil (88% yield). Compound 8
was quantitatively converted into [G1]-C3H6-SiMe2Cl (9) by
hydrosilylation in the presence of a large excess of HSiMe2Cl
using [(NBu4)2PtCl6] as catalyst. Treatment of 9 with an excess
of Li-NCN11a afforded the extended dendrimer ligand [G1]-
C3H6-SiMe2-NCN (10) as a yellow, vicous oil in 87% yield.
Finally, the nickelated derivative [G1]*-Ni12 (11) was prepared
(50% yield) by a similar lithiation/transmetalation procedure
as described earlier for 4, 5, and 7. The moderate yield of 11 is
probably a result of the increased solubility behavior of this
compound. This rendered the purification and workup of 11
less efficient. The dendritic materials 8-11 were identified by
1H, 13C{1H}, and 29Si{1H} NMR spectroscopy, as well as by
MALDI-TOF mass spectrometry and combustion analyses.
A second approach to increase the separation between the
catalytic sites is depicted in Scheme 5 and was aimed at the
introduction of less [G1] periphery-bonded nickel(II) complexes.
This leads to the [G1] nickelated dendrimer 15 (abbreviated as
[G1]-Ni8), which was prepared in three consecutive steps. The
carbosilane dendrimer species [G1]-(CH2CHdCH2)8 (12) was
prepared using a procedure described in the literature4 and fully
characterized17 before its use in subsequent syntheses.
The first step in the synthesis of [G1]-Ni8 (15) was the
hydrosilylation of 12 in neat HSiMe2Cl under platinum catalysis
as described for 9 (vide supra) which gave dendrimer [G1]-
(SiMe2Cl)8 (13: 95% yield). The latter was immediately
converted into the air stable dendrimer [G1]-(NCN)8 (14) by
treatment with an excess of Li-NCN as described for 10. The
nickelated dendrimer 15 was obtained as an orange solid in 79%
yield by using a similar lithiation/transmetalation sequence as
was used for 11. The structures of 12-15 were confirmed by
NMR spectroscopy (1H, 13C{1H}, and 29Si{1H}), mass spec-
trometry (MALDI-TOF), and elemental analyses.
As for the other nickelated dendrimers, an incomplete met-
alation was observed for 11 and 15. The percentage of nickel
incorporation was estimated by 1H NMR spectroscopy and ele-
mental analyses and indicated a nickel loading for both den-
drimers (11: 79%; 15: 89%) falling in the same range as was
established for 4, 5, and 7 (viz. 80-90%). These results clearly
demonstrated the limitations but also the reproducibility of the
lithiation/transmetalation sequence for introducing metal sites
into this type of NCN-containing dendrimer.
a Reagents and conditions: (i) 2 equiv t-BuLi, Et2O, -78 °C; (ii)
0.5 equiv Me2SiCl2, Et2O, -78 °C f RT; (iii) excess Me3SiCl, Et2O,
78 °C f RT; (iv) t-BuLi, pentane, RT; NiCl2(PEt3)2, Et2O.
2 and 3 were synthesized in moderate to good yields (32-86%)
by the reaction of their polylithiated precursors with an
appropriate amount of [NiCl2(PEt3)2] at RT as described below.
The synthesis of the metallodendrimers [G0]-Ni4 (4), [G1]-
Ni12 (5), and [G2]-Ni36 (7) (Schemes 2 and 3) began with the
preparation of the polylithiated precursors of the dendritic ligand
precursors [G0]-SiMe2-NCN,11a [G1]-SiMe2-NCN11a and [G2]-
SiMe2-NCN (6). The second generation NCN-functionalized
carbosilane dendrimer 6 could be prepared in a moderate yield
(48%) as an extremely viscous oil by treatment of the poly-
chlorosilane [G2]-SiMe2Cl4a with an excess of Li-NCN as
described for the [G0]- and [G1]-generation derivatives (vide
supra).11a The polylithiated reagents were obtained as amor-
phous, insoluble red solids by addition of excess of t-BuLi to
these ligand systems and were subsequently purified by remov-
ing the excess t-BuLi by washing with dry hexanes. After this
step, the transmetallating reagent (i.e., [NiCl2(PEt3)2]) was added
as a diethyl ether solution which afforded the multimetallic
dendrimers 4, 5, and 7 after appropriate workup.
To no surprise, the use of these extremely moisture-sensitive
lithium reagents typically resulted in partial hydrolysis during
workup. As a result, we found that nickelation of the ligand
sites beyond 80-90% could not be achieved.17 Spectroscopic
evidence for this is found in the 1H and 13C{1H} NMR spectra
of [G0]-Ni4 (4), [G1]-Ni12 (5), and [G2]-Ni36 (7), in which
signals that were attributed to nonmetalated NCN fragments,
were clearly visible. A single resonance pattern is found for
1
both metalated as well as non-nickelated NCN sites. The H
NMR spectroscopic signal integration of all NCN sites cor-
responded with the signal integration for the SiMe2 groups of
the CS-dendritic support. The microanalyses of 4, 5, and 7 were
in agreement with the calculated percentage of metalation as
1
obtained by H NMR spectroscopy (i.e., signal integration). It
(16) We have found that similar transmetalation reactions with PtII salts
gave comparable results.12b
(17) To our knowledge, no analytical data was available for this particular
dendrimer.
Characterization of (Nickelated) Dendrimers: NMR Spec-
troscopy. As for the related model compounds 2 and 3,18 NMR

Carbosilane-Supported Arylnickel(II) Catalysts
J. Am. Chem. Soc., Vol. 122, No. 49, 2000 12115
Scheme 2: Synthesis of [G0] and [G1] Dendrimers 4 and 5a
a Reagents and conditions: (i) t-BuLi, pentane, RT; NiCl2(PEt3)2, Et2O.
(1H and 13C{1H}) spectroscopic characterization of 4-1518 (C6D6;
RT) revealed single resonances patterns for the CH2N, N(CH3)2,
and Si(CH3)2 groups. This result suggests free rotation for both
the dendritic branches around the central Si (core) atom and
the ligands located on the dendrimer periphery. Of particular
note is that in the 13C{1H} NMR spectrum of 5, the inner and
periphery CH2 fragments of the carbosilane skeleton gave rise
to five separate resonances, while a theoretical sixth resonance
is most probably coincident with one of the signals observed
for the other methylenic groupings. In the 75 MHz 13C{1H}
NMR spectrum obtained for the [G2] dendrimer ligand 6;
however, the signals corresponding for the methylenic carbon
atoms are overlapped at this field, and as a result, only three
broad peaks were observed. This suggests that the intramolecular
rotational freedom in these dendrimer molecules decreases with
increasing generation number. 29Si{1H} NMR spectroscopy
proved to be a very helpful technique for a further structural
assignment of 6. This NMR spectrum showed four (expected)
resonances at 1.48, 1.14, 0.75, and -3.76 ppm, respectively.
Strangely, one additional resonance (δ ) -3.07 ppm) is
observed, and this could indicate the presence of structural
defects present in 6. This latter resonance is ascribed to a silicon
center which is bonded to an aryl group. The observation of
this extra resonance could be caused by the introduction of, for
example, Markovnikov addition of SiMe2Cl groups, followed
by arylation of the resulting â-SiCl groups. We estimated this
defect to be present in only small amounts (∼4%) by means of
signal integration.
separated resonances for the methylenic groups of the carbosi-
lane support of which three were significantly smaller in
intensity. The 29Si{1H} NMR spectrum of the same compound
showed three distinctive resonances (δ ) 1.10, 1.04, and 0.89
ppm, respectively). The hydrosilylation of 8 to give [G1]-C3H6-
SiMe2Cl (9) could be easily monitored, as the resonances of
the olefinic groups in the 1H and 13C{1H} spectra of 8
disappeared with time. This was accompanied by the appearance
of additional resonances assigned to new CH2 and Si(CH3)2
groups. The 29Si{1H} NMR spectrum of 9 clearly showed the
new signal (δ ) 30.76 ppm) attributed to the introduced SiMe2-
Cl fragment. The NMR spectra recorded for [G1]-C3H6-
SiMe2-NCN (10) and [G1]*-Ni12 (11) showed an increased
overlap between the different dendrimer segments and hence
the presence of broader resonance patterns. Similar NMR
spectroscopic observations were encountered for the dendrimer
ligand precursors 12-14 and [G1]-Ni8 (15).
Mass Spectrometric Analysis. Although all nickelated
dendrimers are sensitive to oxidation, one can readily perform
mass spectrometric analyses for the model derivatives 2, 3, and
4. For this purpose, fast atom bombardment (FAB) mass
spectrometry proved to be an effective method for analyzing
these relatively small molecules. This procedure afforded spectra
containing many ion clusters with well-resolved isotopic pat-
terns. The FAB-MS spectra of 2-4 displayed characteristic
(18) The NMR spectra of the (nickelated dendrimer) species tend to suffer
from line-broadening which is most likely caused by the presence of traces
of paramagnetic NiIII. To overcome this problem, NMR samples containing
these compounds can be treated with CO(g) to reduce traces of Ni(III) prior
to spectroscopic measurement which reduces the line-broadening. See:
Grove, D. M.; van Koten, G.; Mul, W. P.; van der Zeijden, A. A. H.;
Terheijden, J. Organometallics 1986, 5, 322-326.
The NMR spectra of the extended dendrimer ligand precursors
8-10 is likewise helpful for structural assignment. Of particular
note is the 13C{1H} NMR spectrum of 8, which displayed six

12116 J. Am. Chem. Soc., Vol. 122, No. 49, 2000
Scheme 3: Synthesis of [G2] Dendrimers 6 and 7a
Kleij et al.
This feature was partially ascribed to an incomplete nickel
incorporation that gives a distribution of dendrimer species, of
which the expected isotopic distributions of the molecular and
fragment ions most likely strongly overlap. The broadness of
these peaks may also be influenced by the oxygen-sensitivity
of these nickelated dendrimers.
We were able to obtain a reasonable mass spectrum for [G2]-
SiMe2-NCN (6) (m/z ) 11453, calcd 11646). The difference
(193) can be explained by an acid-mediated cleavage of one
NCN-H ligand precursor (calculated mass 192) ligand from
the [G2] periphery and the resultant formation of a silicon cation.
Also for [G1]-Ni8 (15) a reasonable MALDI-TOF-MS analysis
could be performed. The mass spectrum obtained for 15 clearly
showed peaks at m/z ) 3446.72, 3409.43, and 3349.42 which
were attributed to the (incompletely nickelated) molecular and
fragment ions [G1]-Ni8Cl8, [G1]-Ni8Cl7, and [G1]-Ni7Cl7.
The mass spectra collected for the precursor compounds and
ligand systems [G1]-SiMe2-CH2CHdCH2 (8), [G1]-C3H6-
SiMe2-NCN (10), [G1]-(CH2CHdCH2)8 (12), and [G1]-
(NCN)8 (14) showed in all cases a clear peak for the molecular
ion consistent with the proposed stoichiometry. It should be
noted that for 8 and 12 a silver salt was added to increase the
peak resolution, which resulted in the observation of (M + Ag)+
molecular ion clusters. In the case of 10 and 14, an acidic matrix
was used to provoke the in situ formation of (M + H)+
molecular ions.
Molecular Modeling of [G0]-Ni4 (4) and [G1]-Ni12 (5).
Molecular mechanics calculations (MM2)19 were performed on
the [G0] and [G1] nickel-containing species to give some
impression about their gas-phase three-dimensional structure.
The size and shape of 5 is shown in Figure 4, which represents
one of the calculated conformations. For 4, a molecular radius
of ∼20 Å was calculated, while for 5 this radius was extended
to ∼30 Å. It should be noted that these idealized structures may
be exaggerated and only give information about the volume of
a fully “stretched” (i.e., spherical) dendrimer. We found a
profound difference between the calculated three-dimensional
structures of 4 and 5. The “front” view of 5 showed this
molecule to be a generalized spherical molecule like in 4. The
“side” view (∼90° rotation) of 5, however, showed a much more
densely packed system. This phenomenon has also been reported
by other authors.20 Nevertheless, this information is useful in
cases where a certain molecular size is prerequisite for the
effective use of ultrafiltration techniques or membrane technol-
ogy for catalyst recovery (vide infra).7,13
While in 4 the Ni sites are well-separated (average Ni-Ni
distance ∼19 Å), in 5 these sites are situated in a much smaller
volume and are in close proximity to each other (closest Ni-
Ni distance about 8-11 Å). It is important to note that the
nonrandom distribution of the nickel complexes in the dendrimer
system 5 could have an effect on the accessibility of the Ni(II)
sites in catalytic applications. In our earlier dendrimer species
(see Figure 1),6 the nickel sites were connected to the dendrimer
periphery via (long) carbamate linkers, and it was anticipated
that the nickel sites are well-separated as in [G0]-Ni4 (4).
Although we were not able to perform molecular modeling for
[G2]-Ni36 (7),21 one could imagine that in this metallodendrimer
the catalytic sites are even more closely packed when compared
to [G1]-Ni12 (5).
a Reagents and conditions: (i) excess Li-NCN, Et2O, -78 °C f
RT; (ii) t-BuLi, pentane, RT; NiCl2(PEt3)2, Et2O.
isotope distributions for their molecular ions (m/z ) 356.2,
626.1, and 1566.2, respectively) while the most intensive peaks
in these spectra were attributed to the [M - Cl]+ fragment ions
(m/z ) 321.2, 591.1, and 1529.1, respectively). Despite the
complexity of these spectra, the presence of molecular ions with
correct isotopic patterns is consistent with the proposed stoi-
chiometry of 2-4. The FAB-MS spectrum of 4 shows clear
isotopic patterns at m/z ) 1472.3, 1437.3, 1378.4, and 1343.4
which are attributed to incomplete nickelated molecular and
fragment ions [G0]-Ni3Cl3, [G0]-Ni3Cl2, [G0]-Ni2Cl2, and [G0]-
Ni2Cl1, respectively. This can, obviously, be a consequence of
incomplete metalation (vide supra) during the synthesis of 4.
We applied MALDI-TOF-MS as an additional analytical tool
for the relatively larger dendrimer molecules 5-15. As could
be expected, the mass spectrometric characterization of the larger
dendrimers was hampered by the appearance of very broad
peaks in the spectra of the nickelated dendrimers 5, 7, and 11.
(19) CAChe Molecular Mechanics at the MM2 level, Oxford molecular
group.
(20) Huck, W. T. S. Thesis Noncovalent Synthesis of Nanosize Metal-
lodendrimers. Technical University of Twente, Twente, 1997.
(21) We were unable to perform Molecular Modeling for the [G2]
derivative 7 with our software (CAChe, MM2) because of the large number
of atoms present in the molecule.

Carbosilane-Supported Arylnickel(II) Catalysts
J. Am. Chem. Soc., Vol. 122, No. 49, 2000 12117
Scheme 4: Synthesis of the Nickelated [G1]* Dendrimer 11a
a Reagents and conditions: (i) allylmmgbr, Et2O, RT; (ii) neat HSiMe2Cl, RT; (iii) excess Li-NCN, Et2O, -78 °C f RT; (iv) t-BuLi, pentane,
RT; NiCl2(PEt3)2, Et2O.
Scheme 5: Synthesis of Nickelated [G1] Dendrimer 15a
voltammetry to investigate the electronic influence of the
dendrimer support on the redox potential of the peripheral [NiCl-
(NCN)] complexes. In principle, these metallodendrimers should
yield cyclic voltammograms that correspond to the Ni(II)/Ni-
(III) redox couple.8a To keep the anion concentration constant
during the oxidation process, an appropiate chloride-containing
salt (viz., [Bu4N]Cl) was used as supporting electrolyte. The
peak potentials for the oxidation (Ep,a) and reduction (Ep,c) were
measured for the model compounds 2 and 3 as well as for the
dendrimer species 4, 5, 7, 11, and 15 and are collected in Table
1. It is important to emphasize that the oxidation process
involved in these Ni(II)Cl complexes is a combination of an
electrochemical as well a chemical reaction (EC-mechanism,
note the newly formed Ni-Cl bond); that is, the transformation
of a square planar Ni(II)Cl complex into a square pyrimidal
Ni(III)Cl2 complex. Consequently, there is no formal Nernstian
relationship between the oxidative and the reductive features.22
Nevertheless, our earlier electrochemical studies with similar
organonickel(II) complexes8a have shown that the oxidation of
the Ni(II) center is (chemically) reversible.
Interestingly, the cyclic voltammograms of all dendrimer
species showed, as in the case of model compounds 2 and 3,
single oxidation and reduction waves consistent with single
potentials in the oxidation and subsequent reduction of the
peripheral nickel-containing complexes. This suggests that the
nickel(II) sites in all dendrimer species are electrochemically
independent. This is in contrast to previously reported ferrocene-
derivatized carbosilane dendrimers.23 Furthermore, the difference
in oxidative and reductive peak potentials (∆Ep) for the den-
(22) Grove, D. M.; van Koten, G.; Mul, P.; Zoet, R.; van der Linden, J.
G. M.; Legters, J.; Schmitz, J. E. J.; Murrall, N. W.; Welch, A. J. Inorg.
Chem. 1988, 27, 2466-2473.
(23) Cuadrado, I.; Casado, C. M.; Alonso, B.; Mora´n, M.; Losada, J.;
Belsky, V. J. Am. Chem. Soc. 1997, 119, 7613-7614.
a Reagents and conditions: (i) neat HSiMe2Cl, RT; (ii) excess Li-
NCN, Et2O, -78 °C f RT; (iii) t-BuLi, pentane, RT; NiCl2(PEt3)2,
Et2O.
Electrochemical Studies. The redox behavior of the orga-
nonickel dendrimers in CH3CN solution was studied by cyclic

12118 J. Am. Chem. Soc., Vol. 122, No. 49, 2000
Kleij et al.
Figure 3. MALDI-TOF-MS of [G1]-Ni8 (15).
Table 1. Cyclic Voltammetry Results of the Model Compounds 2
and 3 and the Nickelated Dendrimers in CH3CNa,b
compound
Ep,a/V
Ep,c/V
∆Ep/V
E1/2/Vc
2
3
-0.103
-0.172
-0.208
-0.278
-0.276
-0.273
-0.298
-0.595
-0.526
-0.435
-0.413
-0.412
-0.414
-0.397
0.492
0.354
0.227
0.135
0.137
0.141
0.099
-0.33
-0.34
-0.32
-0.35
-0.34
-0.34
-0.34
4, [G0]-Ni4
5, [G1]-Ni12
7, [G2]-Ni36
11, [G1]′-Ni12
15, [G1]-Ni8
a For conditions see Experimental Section. b Peak potentials are given
against the Fc/Fc+ couple in the same solvent. c Non-Nernstian midpoint
1
potential defined as E1/2 ) /2(Ep,a + Ep,c).
Catalysis with the Nickelated Dendrimers. In comparative
studies, the arylnickel complexes of general formula [NiX-
(C6H2{CH2NMe2}2-2,6-R-4)] showed to be efficient catalysts
in atom-transfer radical addition (ATRA) chemistry (i.e.,
polyhalogenated alkane addition to olefins, the Kharasch addi-
tion reaction).8 In addition, we recently prepared catalytically
active nickelated carbosilane dendrimers6 with activities per
nickel atom comparable to those observed for their mononuclear
analogues. We therefore investigated the activity of the novel
nickelated dendrimers (vide supra) in the ATRA reaction using
methyl methacrylate (MMA) as substrate and carbon tetrachlo-
ride (CCl4) as the polyhalogenated alkane. The same conditions
were used as reported in our previous studies on this type of
catalysis.6,8 The results obtained for the nickelated dendrimers
were compared with the parent model [NiCl(C6H2{CH2NMe2}2-
3,5-(SiMe3)-4] (2) and di-nickelated Me2Si[C6H2{CH2NMe2}2-
3,5-(NiCl)-4]2 (3). The catalytic data are collected in Table 2
and visualized in Figures 5 and 6.
Figure 4. Space filling model of the “front” and “side” view of the
calculated structure of [G1]-Ni12 (5).
drimer species is significantly smaller than the respective
difference for 2 and 3, whereas the non-Nernstian midpoint
potentials (E1/2, see Table 1) are similar for these species. The
similarity of the measured E1/2 values points to a comparable
electronic influence of the dendritic support and the SiMe3 group
on the nickel center in the dendrimer species and the model
compound 2, respectively.
The total turnover number (TTN) after 4 h, calculated for
the mono-nickel complex 2 was in the same range as that found
for similar [NiX(NCN)] complexes, which were applied in the
same catalytic reaction. The initial reaction rate for 2 reflected
in turnovers per h, however, was approximately 30% lower than
that measured for [NiBr(C6H3{CH2NMe2}2-2,6)], the model
compound used as a standard in our earlier work in this area.6

Carbosilane-Supported Arylnickel(II) Catalysts
J. Am. Chem. Soc., Vol. 122, No. 49, 2000 12119
Table 2. Catalytic and Kinetic Data for the ATRA Reaction of
CCl4 to MMA Using the Model Compounds 2 and 3 and the
Nickelated Dendritic Catalystsa
time
(h)
conversionb
(%)
TOF
TTN
(per Ni)
compound
(per Ni per h)c
2
3
0.25
2
0.25
2
0.25
2
0.25
2
0.25
2
0.25
2
0.25
2
15
91
9
74
9
79
3
17
0.2
1.0
8
163
108
111d
44d
3d
277e
311e
324d,e
63d,e
4, [G0]-Ni4
5, [G1]-Ni12
7, [G2]-Ni36
11, [G1]*-Ni12
15, [G1]-Ni8
11d,e
102d
53d
310d,f
284d,f
Figure 7. Display of the retained fractions of [G0]-Ni4 (4) and [G1]-
Ni12 (5) in a membrane reactor.26
54
5
27
substituent R (Figure 2) that has the greatest influence on the
catalytic activity. This latter effect has been studied and reported
elsewhere.8a The halide influence on the initial reaction rate is
known to decrease in the order I ≈ Br > Cl.8a
a In all catalytic runs (duplo experiments) the concentration of nickel
was kept constant at ∼9 × 10-5 mol. b Selectivity for the formation of
the 1:1 adduct of methyl methacrylate and CCl4 is >99%. Conver-
sion calculated from the formation of the 1:1 addition product as
determined with GC analysis using dodecane as an internal standard.
c During the first 15 min. d Values corrected for nickel incorporation
using the NMR spectroscopic and elemental analyses data. e After 4 h.
f After ∼22 h.
The decrease in catalytic activity found for 4 compared to 2
in terms of the initial reaction rate (i.e., TOF/h during the first
15 min, Table 2) is ∼30% which is in line with our earlier results
in this area.6 For [G1]-Ni12 (5) however, a dramatic drop of
∼70% in initial catalytic activity per nickel atom was observed
together with the precipitation of an insoluble purple/brown solid
(P-5) after ∼1 h. The large difference in (initial) activity and
TTN/Ni site between 5 and the model derivatives 2 and 3 could
not be explained by the fact that 5 is not fully metalated.
Additionally, for [G2]-Ni36 (7) almost no catalytic activity
(conversion <1.5% after 4 h) was observed. Furthermore, after
∼3 h, again an insoluble purplish solid (P-7) precipitated which
did not exhibit any catalytic activity (Figure 5).
The two alternative [G1] nickelated carbosilane dendrimers
[G1]*-Ni12 (11) and [G1]-Ni8 (15) were also applied in the same
standard ATRA reaction (Figure 6 and Table 2). A marked
difference was observed when the catalytic activity of 5 was
compared with the extended analogue [G1]*-Ni12 (11). First,
the conversion of MMA into the addition product after 4 h was
already 49% (cf., the maximum conversion of ∼20% in the case
of 5) and became almost quantitative (97%) after 22 h when
11 was used as catalyst. A second major difference is that the
precipitation of a purple solid did not occur until all MMA had
been converted into product. This latter phenomenon was
monitored in time, and after 6 h, still no observable precipitation
had occurred. It must be emphasized that the initial reaction
rates per nickel site in 11 and 4 were comparable. The TTN for
11 (310) was comparable with those obtained in the catalytic
reactions employing the model derivative 2 or 4 with TTNs of
277 and 324, respectively. The use of the nickelated dendrimer
[G1]-Ni8 (15) (TTN ) 284) gave results similar to those of
[G0]-Ni4 (4) and [G1]*-Ni12 (11), with (only) a slightly reduced
initial reaction rate (Table 2).
Figure 5. Visualization of the catalytic results obtained for nickelated
compounds 2, 4, 5, and 7. Abbreviations used: C ) conversion, t )
time. The curve corresponding to the catalytic performance of 3 has
been omitted for clarity reasons.
Membrane Catalysis with [G0]-Ni4 (4) and [G1]-Ni12 (5).
Dendrimers 4 and 5 were also tested for their degree of retention
(Figure 7)24 and catalytic characteristics (Figure 8) in a mem-
brane reactor equipped with a SelRO-MPF-50 nanofiltration
membrane.25 Although the measured retention in this membrane
system was neither quantitative for 4 (97.4%) nor for 5
(99.75%), theoretical calculations (Figure 7) showed that 5
Figure 6. Visualization of the catalytic results obtained for nickelated
compounds 5, 11, and 15. Abbreviations used: C ) conversion, t )
time.
This lower initial activity was attributed to the presence of a
Ni-Cl bond in 2 (cf., Ni-Br bond in [Ni(NCN)Br]). As there
is a fast exchange of the bromide for a chloride anion in the
unsubstituted complex [Ni(NCN)Br] in the presence of a large
excess of CCl4 during catalysis,8a,b it is eventually the para
(24) The degree of retention for [G0]-Ni4 (4) and [G1]-Ni12 (5) in CH2-
Cl2 at 25 °C using the SelRO-MPF-50 membrane was measured by
photometric measurements taking into account both the nickel concentration
in the retained fraction and in the “leached” fraction.

12120 J. Am. Chem. Soc., Vol. 122, No. 49, 2000
Kleij et al.
Figure 9. Continuous membrane catalysis carried out with 5. Ab-
breviations used: STY ) space time yield, t ) time.
Figure 8. Display of the membrane catalysis with 4 and 5. Abbrevia-
tions used: C ) conversion, t ) time.26
(Figure 9) with a continuous feed of [Bu4N]Br to prevent catalyst
precipitation.
Table 3. Membrane Catalysis Results Obtained for 4 and 5a
time
(h)
conversion
(%)
TOF
(per Ni per h)
TTN
(per Ni)
Despite the fact that 5 was successfully retained in the reactor
and that a precipitation of 5 was avoided, a fast decrease of the
conversion per time unit (after 33 h ≈ 0) was observed with a
low TTN/Ni of only 20. After the continuous membrane reaction
(i.e., after the reactor volume was replaced 64 times), both the
retained fraction as well as the leached fraction were subjected
to ICP mass spectrometric analysis in order to calculate the
amount of nickel present. From these data, the degree of reten-
tion for 5 in the continuous operating membrane reactor system
was determined at 98.6%, which closely resembled the measured
retention found for 5 (99.8%, vide supra) in a batch reactor.
Although a retention of 98.6% theoretically leads to loss of cat-
alytic material relatively quickly inside the continuous mem-
brane reactor, it cannot solely explain the observed fast deac-
tivation of [G1]-Ni12 (5) in the reactor.
compound
4, [G0]-Ni4
0.25
20.4
0.25
23
0.25
91.4
11
74
9.8
60
7
98c
88c
88c
173c
154c
148c
5, [G1]-Ni12
5, [G1]-Ni12
b
45
a See Table 2 for details of this catalysis. b In the presence of a large
excess of [NBu4]Br. c Values corrected for nickel incorporation using
the NMR spectroscopic and elemental analyses data.
should be sufficiently large enough for the use in a continuous
operating reactor without significant leaching of the catalyst after
100 cycles.26 The leaching of the catalytic species not only will
depend on the membrane perfection and permeability, but also
on the chemical stability of the catalyst system. It is therefore
Mechanistic Arguments. In the ATRA (i.e., Kharasch)
reaction of CCl4 to MMA, the addition of a nickel catalyst to a
mixture containing these substrates is accompanied by an
immediate color change, giving rise to purple/brown solutions.
For the model catalysts 2 and 3 and dendrimer species 4, 11,
and 15, purple precipitates started to form after the conversion
of MMA were virtually complete. In contrast to this observation,
when 5 or 7 were used as catalyst in this system, the pre-
cipitation of an insoluble purple solid and a total deactivation
of the catalytic system was already noticed after approximately
1 and 3 h, respectively, while conversion of MMA to the 1:1
addition product of only 18 and 1.5%, respectively, was
achieved. Similar observations of the formation of insoluble
purple species during catalysis were made in the membrane
experiments carried out with 4 and 5.
Interestingly, comparable observations were performed for
polymer-bound [Ni(NCN)Br] complexes applied in ATRA
chemistry.28 Spectroscopic studies (ESR, FT-IR, and UV/Vis)
indicated that the purple compound in this case was a material
containing paramagnetic Ni(III) centers. Furthermore, the
catalytic activity of these polymer catalysts was strongly
dependent on the nickel catalyst loading. For the polymeric
catalytic materials with a relatively low loading of nickel catalyst
(cf., [G0]-Ni4 (4)), the precipitation of a purple/brown solid
during catalysis did not appreciably lower the activity of the
immobilized Ni(II) complexes. A high increase in nickel loading
in the same polymer system (cf., [G1]-Ni12 (5) and [G2]-Ni12
(7)) led to a significantly lower catalytic activity per nickel atom
(∼60%) attributed to a decrease in initial solubility of the
catalytic species. The latter feature was held responsible for the
important to note that the carbon-metal bond (i.e., Ni-Caryl
and the connecting dendrimer-ligand Si-C bonds in 4 and 5
are not easily cleaved under the reaction conditions.
)
During the catalytic membrane experiments in the presence
of 4 or 5, the precipitation of insoluble purple species (P-4 and
P-5) after 40 min. was again noted. However, in contrast to the
standard batch reaction carried out with [G1]-Ni12 (5), P-5 still
showed significant catalytic activity in the Kharasch reaction,
and in the case of P-4, conversion of MMA into the adduct
was achieved up to 74% after ∼20 h (Table 3, Figure 8). The
addition of a large excess of [Bu4N]Br did, however, prevent
the formation of P-5 in the case of 5 (Table 3), but as a result,
part of the addition product contained both chlorine and bromine
atoms.27 Another effect from the addition of [Bu4N]Br is a lower
activity (after 15 min) when compared to the reaction without
this additive. After ∼90 h, the conversion reached only 45% in
the presence of 5 as catalyst. These latter results significantly
differed from those obtained for 5 in the batch reaction (Table
2) and pointed to a slower deactivation process. We therefore
attempted to use [G1]-Ni12 (5) in a continuous operating mem-
brane reactor system equipped with a SelRO-MPF-50 membrane
(25) A SelRO-MPF-50 membrane was used and this membrane consists
of polyphenylen- oxide-, polysulfonated-, and haloalkylated polyphenele-
noxide-based materials. This membrane is hydrophobic and can retain
opportunistic H2O. The reactor material was made from poly(propylene).
(26) A cycle or residence time is defined as the time needed to fully
replace one membrane reactor volume.
(27) This has also been observed in crossover experiments with [NiBr-
(NCN)] in the presence of MMA and a mixture of CCl4/CBr4. The
experiments revealed the formation of eight products (all 1:1 adducts). This
has been ascribed to a statistical incorporation of both bromide as well as
chloride into the products through the initial formation of an “inner- sphere”
intermediate.9 The [NBu4]Br additive is thought to be involved in a halide
methathesis reaction with the initially formed Ni(III)Cl2 intermediate.
(28) van de Kuil, L. A.; Grove, D. M.; Zwikker, J. W.; Jenneskens, L.
W.; Drenth, W.; van Koten, G. Chem. Mater. 1994, 6, 1675.

Carbosilane-Supported Arylnickel(II) Catalysts
J. Am. Chem. Soc., Vol. 122, No. 49, 2000 12121
Scheme 6: Postulated Key Intermediate Steps in the ATRA
(i.e., Kharasch) Reaction Catalyzed by [NiCl(NCN)]
Complexesa
Figure 10. Displacement ellipsoid plot (ORTEP, 50% probability level)
of the molecular structure of 16. Selected bond lengths (Å) and
angles (deg) with esd’s in parentheses: Ni-Cl1 2.2761(5), Ni-Cl2
2.2756(6), Ni-C9 1.8850(16), Ni-N1 2.0324(14), Ni-N2 2.0283(14),
Cl1-Ni-C9 164.01(6), N1-Ni-N2 154.05(6), N1-Ni-C9 82.46-
(6), Cl1-Ni-N1 93.44(4), Cl2-Ni-N1 100.34(4).
fundamental steps in the ATRA (i.e., Kharasch) reaction
catalyzed by metal halide complexes. Similar mechanistic steps
are also postulated in Scheme 6. The abstraction of a halide
atom from CCl4 and the concomitant inner-sphere oxidation of
the Ni(II) complex produces a persistent radical that does not
self-terminate or propagate (i.e., the newly formed Ni-Cl bond
is not activated). The transient CCl3 radical, however, is involved
in both termination as well as addition reactions. In general,
the highly reactive character of this transient radical leads to a
very low steady-state concentration of radicals, and thus the
contribution of termination is, therefore, negligible. Another
important aspect is that the polyhalogenated substrate should
generate radicals more easily than the (final) products and should
only give monoaddition. It was found that the catalytic system
MMA/CCl4/[NiX(C6H2{CH2NMe2}2-3,5-R-4)] (Figure 2) in-
deed obeys all of these characteristics put forward by the
persistent radical concept, and this makes it possible to give an
explanation for the (membrane) catalysis results and the
observed precipitation of purple solids during catalysis (vide
infra).
a Abbreviations used: kf ) forward reaction, kb ) back reaction, kp
) forward product-forming reaction, kr ) back “radical-reforming”
reaction.
formation of a relatively inactive, “heterogeneous” reaction mix-
ture. The reduced solubility characteristics found for 5 and 7
under standard homogeneous conditions and the (total) deactiva-
tion of these catalysts are in line with these former results.
The purple color observed in this type of ATRA reaction
catalyzed by [NiX(NCN)] complexes was associated with the
formation of a relatively stable, intermediate species in the
postulated catalytic cycle (Scheme 6).9 In the first step of this
cycle, the nickel(II) complex reacts with a polyhalogenated
alkane, in this case CCl4, to give a nickel(III) intermediate (i.e.,
a persistent radical, vide infra) and a CCl3 transient radical (see
eq 1, Scheme 6). The latter stays fixed in the coordination sphere
of the Ni(III) center.29 In a subsequent step (eq 2), the CCl3
radical reacts with the olefinic substrate to yield a product radical
P•, which in the last step (eq 3) is converted into a perhaloge-
nated product by an atom-transfer “back” reaction from the
persistent radical (i.e., the Ni(III) complex) to regenerate the
catalytically active Ni(II) species. Although this type of catalysis
involves highly reactive transient radical species, the selectiVity
for the ATRA reaction catalyzed by [NiX(NCN)] complexes is
virtually quantitative when CCl4 is used in a large excess.30
Identification of the Purple Species. To identify the
insoluble purple species that precipitated during catalysis under
the standard batch conditions, we performed a model studies
with the mononuclear model [NiCl(C6H2{CH2NMe2}2-2,6-
SiMe3-4)] (2). Thus, the organonickel(II) complex 2 was treated
with an excess of CCl4 in the absence of MMA (eq 2).
The observed selectivity behavior can be explained by the
persistent radical effect, that was recently formulated and
reviewed by Fischer.31 At the same time, Matyjaszewski32 used
the concept of the persistent radical effect to describe the
(29) When a mixture of CBr4/CCl4 and [NiBr(NCN)] is analyzed with
GLC/GC-MS, three organic products were identified as CBrCl3, CBr2Cl2,
and CBr3Cl. This corroborates the view that the initial process comprises
the formation of an inner-sphere activated complex derived from the nickel
complex and the polyhalogenated alkane.
(30) Under the ATRA conditions employed (i.e., a high polyhalogenated
alkane/Ni catalysts ratio) there is exclusive formation of the 1:1 adduct.
However, at a low concentration of this polyhalogenated alkane, controlled
radical polymerization of MMA can be carried out with this type of nickel
complexes. See: Granel, C.; Dubois, Ph.; Je´roˆme, R.; Teyssie´, Ph.
Macromolecules 1996, 29, 8576.
(31) Fischer, H. J. Polym. Sci., Part A: Polym. Chem. 1999, 37, 1885
and references therein.
(32) (a) Matyjaszewski, K., Ed. Controlled Radical Polymerization; ACS
Symposium Series 685, American Chemical Society, Washington, DC, 1998.
(b) Matyjaszewski, K. Chem. Eur. J. 1999, 5, 3095 and references therein.
This reaction was carried out to mimick the termination of
ATRA catalysis by [NiX(NCN)] complexes (i.e., when [MMA]
≈ 0). The addition of CCl4 to 2 leads immediately to a dark
purple to brown solution. After workup and crystallization of
the product, dark red to brown crystals were obtained, which
were analyzed by microanalyses and pointed to the formation
of the corresponding Ni(III) dichloride species [NiCl2(C6H2{CH2-
NMe2}2-2,6-SiMe3-4)] (16). To unambiguously establish the
structure of the product, an X-ray crystal structure determination
was carried out. The molecular structure of 16 is depicted in
Figure 10 (see the Supporting Information for further details).
Figure 10 shows that the geometry around the formal Ni(III)

12122 J. Am. Chem. Soc., Vol. 122, No. 49, 2000
Kleij et al.
give rise to a signal with a relative high dipolar broadening
(Figure 11B). In the other phase, the Ni(III) sites are more
separated and this “spectrum” is comparable to the ESR
spectrum of 16 (Figure 11A) measured at low temperature.
Combination of the latter two spectra yielded the theoretical
spectrum C, which is similar to the measured ESR spectrum
for P-5 (Figure 11D).35
Two stoichiometric mixtures of [NiIICl(C6H2{CH2NMe2}2-
2,6-SiMe3-4)] (2) and [NiIIICl2(C6H2{CH2NMe2}2-2,6-SiMe3-
4)] (16), and 2 and [NiIIIBr2(NCN)] (i.e., Ni(II)/Ni(III) mixed
valence complex mixtures) were also subjected to ESR spec-
troscopic investigations in toluene glass at ∼120 K. The ESR
spectra obtained for both mixtures were identical to the ESR
spectra of the single Ni(III)X2 (X ) Cl or Br) species gathered
under the same conditions and no clear evidence was found for
a significant intermolecular interaction between the Ni(II) and
Ni(III) complexes under these experimental conditions.
Mode of Deactivation. In the absence of olefinic substrate
(i.e., when [MMA] ) 0), the Ni(III) derivative [NiCl2(C6H2{CH2-
NMe2}2-2,6-SiMe3-4)] (16) is produced by a process that
probably involves a C-Cl bond activation of CCl4 by the Ni-
(II) complex 2 (eq 2). The formation of C2Cl6 is obvious
evidence for this and is ascribed to the termination reaction of
two CCl3 radicals. For each C2Cl6 molecule formed in this way,
2 equiv of the persistent radical 16 remain which in this case
cannot be recycled into the Ni(II) complex. To no surprise, in
the presence of a cosolvent, another organic product (i.e., C3H2-
Cl6) is formed. The formation of the latter is likely a result of
a reaction of the transient radical CCl3 with the chlorinated
solvent. This suggests that the transient radical is involved in
outer-sphere reactions, and, furthermore, that the reaction
behavior of the transient radical is no longer controlled by the
persistent radical (i.e., 16).
Molecular modeling performed on 5 indicated that the Ni
sites in this dendrimer are in much closer proximity when
compared to 4 and the nickelated [G1] dendrimer described
earlier (Figure 1).6 In this regard, it should be noted that in these
dendrimer species the [NiCl(NCN)] catalytic sites are nonran-
domly distributed. During catalysis, this latter feature may lead
to proximity effects such as termination (via coupling as
observed in the synthesis of the Ni(III) complex 16) of radical
intermediates or electron-transfer processes within the dendrimer
molecule itself. Obviously, these (kinetically based) deactivation
processes will likely depend on the spatial separation (or
proximity effect) of the Ni(II) catalytic sites. This was supported
by the use of the two alternative [G1] dendrimer catalysts [G1]*-
Ni12 (11) and [G1]-Ni8 (15), in which the Ni(II) sites are more
separated from each other as compared to [G1]-Ni12 (5) and
[G2]-Ni36 (7). These data supported the view that the effective-
ness of the back reaction between the peripheral Ni(III)
persistent radical and the product radical P• (Scheme 6) follows
the series 7 < 5 < 2. Ultimately, outer-sphere reactions of the
transient radical species CCl3 as described in the preparation
of 16 would give rise to irreVersible formation of catalytically
inactiVe Ni(III) complexes on the periphery of the dendrimer.
The formation of abundant Ni(III) sites on the periphery of the
dendrimer was confirmed for the precipitate P-5 obtained from
dendrimer 5. The solid-state ESR spectrum of P-5 clearly
showed a large resemblance with the ESR spectrum recorded
Figure 11. ESR spectra recorded for the nickel(III) complex 16 and
the precipitate P-5 derived from [G1]-Ni12 (5).
center is best described as square pyramidal with the nickel atom
displaced from the basal plane. No unusual bond distances and
bond angles were found for 16 when compared to a related
organonickel(III) diiodide complex.33
We further investigated the formation of 16 by analysis of
the organic products (GC-MS) when 2 was exposed to an
excess of CCl4 in the absence of MMA. From the mass spectra
it became clear that two major organic products were obtained
depending on the presence of a cosolvent (i.e., CH2Cl2). In the
presence of CH2Cl2, two products were identified with fragment
ions that displayed peaks at m/z ) 201 and 215. The former
peak could easily be assigned to the fragment ion C2Cl5+ while
the latter peak could be ascribed to a fragment ion with a
molecular formula C3H2Cl5+. The presence of characteristic
isotope patterns due to the presence of multiple chloride atoms
in both ions suggested the formation of the perhalogenated
alkane compounds C2Cl6 and C3H2Cl6. The structure of the latter
compound was not elucidated. It should be noted that in the
absence of CH2Cl2, only the formation of C2Cl6 was observed
(eq 2).34
The purple precipitate P-5 obtained from the dendrimer
catalyst 5 using the standard catalytic protocol was also isolated
and a mass balance of ∼80% was calculated taken into account
a full conversion of the Ni(II)Cl into Ni(III)Cl2 sites in 5. The
purple precipitate P-5 and the model derivative [NiCl2(C6H2{CH2-
NMe2}2-2,6-SiMe3-4)] (16) were analyzed by means of ESR
spectroscopy (vide infra).
ESR Spectroscopy of P-5 and 16. Due to the insolubility
of the purple/brown precipitate P-5 in all the common organic
solvents and H2O, we used solid-state ESR spectroscopy for
an analysis. The ESR spectrum obtained for P-5 (Figure 11D)
was compared to the spectrum obtained for the parent Ni(III)
compound [NiCl2(C6H2{CH2NMe2}2-2,6-SiMe3-4)] (16) in tolu-
ene glass at 120 K (Figure 11A).
The solid-state ESR spectrum of P-5 largely resembles the
ESR spectrum of 16, and the former is regarded as a superposi-
tion of independent spectra for different Ni(III) complexes.
Theoretical modeling showed that spectrum D most probably
consists of two phases. In one phase, the nickel(III) sites are
more closely packed, leading to a faster relaxation mode and
(33) Grove, D. M.; van Koten, G.; Zoet, R.; Murall, N. W.; Welch, A.
J. Am. Chem. Soc. 1983, 105, 1379.
(34) The quantitatiVe analysis of the formation of C2Cl6 by treatment of
the nickelated derivative with CCl4 was hampered because of the insoluble
character of the nickel complex in this solvent. The same accounts for the
dendrimer species. In the presence of a co-solvent, another product with
unknown structure (probably C3H2Cl6) was found as mentioned in the main
text.
(35) We attempted a spin quantification in P-5 using 16 and a CuSO4
sample as standards. The results indicated a (low) conversion of 13% of
the present Ni(II) into Ni(III) sites. However, due to the nonrandom
distribution of the Ni(II) sites in P-5 compared to 16, this figure may be
speculative. As for 16, the same g-values were found for P-5 with an
increased line-broadening.

Carbosilane-Supported Arylnickel(II) Catalysts
J. Am. Chem. Soc., Vol. 122, No. 49, 2000 12123
nanofiltration membranes which can stand long-term exposure
to (chlorinated) organic solvents will play a crucial role.
Conclusions
In summary, we have detailed a new and simple synthetic
methodology for the production of metal-centered carbosilane
dendrimers by using polylithiated precursors. These new met-
allodendrimers were successfully employed as homogeneous
catalysts in the ATRA (i.e., Kharasch) reaction. The use of
ultrafiltration membrane technology as a tool for the separation
of macromolecular catalysts from the product stream was
demonstrated. The size properties of [G1]-Ni12 (5) (∼25-30 Å
diameter) allowed the catalysis to be operated under continuous
conditions with retention of the catalyst in the membrane reactor
and with no significant loss of catalytic material. However, the
formation of a purple precipitate, which contains Ni(III) sites,
was a limiting factor. From this work it became clear that there
exists a “proximity effect” between the peripheral Ni(II) sites
in [G1]-Ni12 (5) and [G2]-Ni36 (7), which translates into lower
catalytic efficiencies and irreversible formation of inactive Ni-
(III) sites. Our future work will be focused on this aspect and
we are currently investigating the synthesis and application of
other types of dendrimer catalysts.
Figure 12. Schematic deactivation mode for dendrimer catalysts [G1]-
Ni12 (5) and [G2]-Ni36 (7).
for the mononuclear model [NiCl2(C6H2{CH2NMe2}2-2,6-
SiMe3-4)] (16). In general, the Ni(III)X(NCN) complexes are
less soluble in the common organic solvents compared to their
respective Ni(II) derivatives. Therefore, the observation of purple
precipitates during these catalytic reactions with 5 and 7 further
supported the proposed formation of polynickel(III) dendrimers.
This mode of deactivation of 5 and 7 is schematically visualized
in Figure 12 and is based on all the presented results and
mechanistic arguments (vide supra).
Experimental Section
General. All sensitive, organometallic manipulations were performed
under a dry and deoxygenated dinitrogen atmosphere using standard
Schlenk techniques unless otherwise stated. All solvents were carefully
dried and distilled prior to use. The reagents that were purchased
commercially were used without further purification. The nanofiltration
membrane SelRO-MPF-50 was obtained from Koch International
GmbH, Du¨sseldorf, Germany. The starting materials [G1]-SiMe2Cl,4a
[G2]-SiMe2Cl,4a [G1]-(CH2-CHdCH2)8,4 [G0]-SiMe2-NCN,12a [G1]-
SiMe2-NCN,11a 1-bromo-3,5-bis[(dimethylamino)methyl]benzene,14
and 1-trimethylsilyl-3,5-bis[(dimethylamino)-methyl]benzene14 were
prepared according to literature procedures. 1H, 13C{1H}, and 29Si{1H}
NMR spectra were recorded on a Bruker AC200/AC300 or Varian
Inova/Mercury 200 or 300 MHz spectrometer. Chemical shifts are given
in ppm using TMS as an external standard. Coupling constants are
given in Hertz (Hz). FAB-MS spectra were obtained from the Analytical
Chemistry Department of the Utrecht University. Photometric measure-
ments were carried out with a Shimadzu UV160 A, CPS-240 A
apparatus. ESR spectra were obtained from the Department of Molecular
Spectroscopy of University of Nijmegen. Elemental analyses were
performed by Dornis und Kolbe, Mikroanalytisches Laboratorium,
Mu¨lheim a.d. Ruhr, Germany. Please note that a full description of the
experimental procedures and analytical data of compounds 1-5 and
8-15 have been deposited as Supporting Information. The procedures
described here for compounds 6 and 7 may be regarded as examples
for the other model/dendrimer compounds.
A distinct difference between the batch reactions and the
membrane experiments was found for the catalytic activity of
the purple precipitates P-4 and P-5. Whereas for 5 a total
deactivation was observed under the batch conditions, the purple
precipitate P-5 in the membrane reactor showed better catalytic
activity and significantly higher substrate conversion (Tables 2
and 3), although the reaction rate was relatively low. It is
possible that the radical intermediates involved in this ATRA
reaction (Scheme 6) interacted with groups present in the
polymeric material of the membrane.36 Consequently, the
irreversible formation of Ni(III) sites in the dendrimer catalyst
5 was slowed down. The addition of [NBu4]Br prevented the
precipitation of the catalytic species on the membrane surface
during catalysis. The perhalogenated products contained, as a
result, some bromine atoms. This indicates that the addition of
halide could compete with the presumed interaction of the
catalytic species with the membrane surface. As a logical
consequence of the addition of an excess of [NBu4]Br the
catalytic activity was relatively lowered (Scheme 6). Neverthe-
less, the membrane catalysis results obtained with 4 and 5 helped
to establish that recycling of these dendrimers is possible. This
work also demonstrates that the compatibility between the
membrane system and the type of catalysis is an important
parameter that, at least in this case, needs to be studied in more
detail and optimized before an efficient continuous catalysis
operation can be carried out. Therefore, the development of other
[G2]-SiMe2-NCN (6). This compound was prepared in a similar way
as has been described for [G0]-SiMe2-NCN and [G1]-SiMe2-NCN.11a
To a solution of 1-bromo-3,5-bis[(dimethylamino)methyl]benzene (3.04
g, 11.2 mmol) in Et2O (30 mL) was added t-BuLi (11 mL of a 1.7 M
solution in pentane, 18.7 mmol) at -78 °C and the resultant white
suspension stirred for 20 min, whereupon [G2]-SiMe2Cl (1.21 g, 0.201
mmol) in Et2O (25 mL) was added. After isolation and purification, a
(36) A deactiVation of the catalytic species by interaction with either
the membrane material or the reactor material26 can be ruled out. This has
been investigated by separate experiments in which the influence of the
membrane material or reactor material has been studied separately. However,
a positiVe influence on the catalytic activity of [G0]-Ni4 (4) and [G1]-Ni12
(5) by such interactions is possible leading to a slower deactivation process
(i.e., Ni(III) formation).
(37) Although we have done several attempts to obtain reliable analytical
(C, H, N) data for 11 and 15, in each case the measured analyses were
significantly lower than expected. The relative ratios, however, do cor-
respond to the expected relative ratios and point the proposed ligand-to-
metal combination.
1
extremely viscous oil was obtained (1.13 g, 0.097 mmol, 48%). H
NMR (C6H6): δ ) 7.63 (s, 72H, Ar-H), 7.58 (s, 36H, Ar-H), 3.46
(s, 144H, CH2N), 2.22 (s, 432H, N(CH3)2), 1.69 (br, SiCH2CH2-), 1.07
(br, -CH2SiMe2-), 0.87 (br, SiCH2CH2-), 0.45 (s, 216H, Si(CH3)2).
13C{1H} NMR (C6H6): δ ) 139.3, 139.0, 133.1, 130.6 (Ar-C), 64.7
(CH2N), 45.6 (N(CH3)2), 21.2, 19.3, 17.9 (3 × CH2), -2.2 (Si(CH3)2).
29Si{1H} NMR (C6D6): δ ) 1.48, 1.14, 0.75 (inner + outer Si), -3.76
(Si-aryl). MALDI-TOF-MS: m/z 11453 (M - NCN)+. Calcd m/z
11454. Anal. Calcd for C660H1212N72Si53: C 68.07, H 10.49, N 8.66, Si
12.78. Found: C 68.17, H 10.38, N 8.61, Si 12.70%.

12124 J. Am. Chem. Soc., Vol. 122, No. 49, 2000
Kleij et al.
[G2]-Ni36 (7). To a solution of 6 (0.29 g, 0.025 mmol) in pentane
(40 mL) was added t-BuLi (2.4 mL of a 1.7 M solution in pentane, 4.0
mmol). The resultant red suspension was stirred for 1 h and the excess
t-BuLi removed by decantation of the pentane layer. Then a solution
of NiCl2(PEt3)2 (0.61 g, 1.67 mmol) in Et2O (40 mL) was added and
the obtained orange to red suspension stirred for 17 h. Orange solid
(0.12 g, 0.008 mmol, 32%). 1H NMR (C6H6): δ ) 6.90 (s, 72H, Ar-
H), 3.37 (s, 144H, CH2N), 2.57 (s, 432H, N(CH3)2), 1.75 (br,
SiCH2CH2-), 1.10 (br, -CH2SiMe2-), 0.95 (br, SiCH2CH2-), 0.49
(s, 216H, Si(CH3)2). MALDI-TOF-MS: m/z 5000-15000 (very broad
peak). Anal. Calcd for 79% nickelated C660H1176N72Si53Ni36Cl36: C
56.05, H, 8.44, N 7.13, Ni 11.13. Found: C 55.65, H 8.28, N 6.22, Ni
11.33%.
[NiCl2(C6H2{CH2NMe2}2-2,6-(SiMe3)-4)] (16). To an orange-
colored solution of 2 (96.8 mg, 0.271 mmol) in CH2Cl2 (25 mL) was
added an excess of CCl4 (∼5 mL). Instantaneously, a dark brownish
solution was obtained, and the mixture was stirred for 16 h. After
evaporating the solvent in vacuo and recrystallization from CH2Cl2/
pentane, dark red to purple crystals were isolated which were suitable
for an X-ray molecular structure determination. Yield: 42.2 mg (0.107
mmol, 40%). ESR (0.5 mg/mL in toluene glass at 150 K): gx ) 2.3879,
gy ) 2.2009, gz ) 2.0261 (hyperfine coupling: 28 G). FAB-MS: m/z
391.1 (M+). Anal. Calcd for C15H27N2SiNiCl2: C 45.83, H 6.92, N
7.13. Found: C 45.79, H 6.96, N 7.06%.
protocol. The applied nanofiltration membrane (SelRO-MPF-50) was
conserved in a aqeous solution of glycerine bisulfite, washed with
acetone and conserved in acetone for 24 h prior to use.
X-ray Crystal Structure Determination of 16. C15H27N2SiNiCl2,
Mr ) 393.09 g/mole, dark red plate, 0.38 × 0.30 × 0.05 mm3,
monoclinic, P21/c (no. 14), a ) 9.9622(2) Å, b ) 12.8152(3) Å, c )
15.5640(3) Å, â ) 106.7933(14)°, V ) 1902.28(7) Å3, Z ) 4, F )
1.373 g/cm3. 13584 reflections were measured on a Nonius Kappa CCD
diffractometer with rotating anode (λ ) 0.71073 Å) at a temperature
of 150(2) K. 4354 reflections were unique (Rint ) 0.0421). Absorption
correction with PLATON38 (multiscan, µ ) 1.359 mm-1, 0.53-0.59
transmission). Structure solved with Patterson methods (DIRDIF9739)
and refined with SHELXL9740 aganist F2 of all reflections. Non-
hydrogen atoms were refined freely with anisotropic displacement
parameters. Hydrogen atoms were located in the difference Fourier map
and refined freely with isotopic displacement parameters. 298 refined
parameters, R-values [I > 2σ(I)]: R1 ) 0.0272, wR2 ) 0.0648. R-values
[all refl.]: R1 ) 0.0392, wR2 ) 0.0694. Molecular illustration, structure
checking and calculations were performed with the PLATON package.38
(See for further details the Supporting Information).
Acknowledgment. We are indebted to Cees Versluis from
the Analytical Chemistry Department of the Utrecht University
for FAB-MS measurements. A.W.K., M.L., and A.L.S thank
the Council for Chemical Sciences (CW) and The Netherlands
Organization for Scientific Research (NWO) for financial
support.
Electrochemical Measurements. All experiments were performed
in an electrochemical cel where a nitrogen atmosphere was maintained.
The measurements were made with a conventional three-electrode
configuration using platinum electrodes and a Ag/AgCl reference
electrode. Cyclovoltammetric studies were carried out in CH3CN
solutions containing the supporting electrolyte (0.1 M [Bu4N]Cl). The
reference electrode was separated from the test solution by a glass frit.
Under the experimental conditions, E1/2 for (Fc/ Fc+) ) +0.053 V (∆Ep
≈ 80-100 mV). An EG&G potentiostat/galvanostat model 263A was
used connected to model 270/250 Research Electrochemistry Software
(version 4.23). Cyclic voltammograms were obtained with a scan rate
Supporting Information Available: A listing of tables of
atomic coordinates, bond lengths and angles, thermal parameters
and relevant crystallographic data for compound 16, Figure 3
(MALDI-TOF-MS analysis of 15), a complete list of experi-
mental procedures and analytical data for compounds 1-5 and
8-15 (PDF). This material is available free of charge via the
Internet at http://pubs.acs.org.
of 100 mV s-1
.
Standard Catalytic Protocol. Prior to use, a mixture of MMA (28
mmol), CCl4 (104 mmol) and dodecane (internal standard, 8.9 mmol)
was degassed thoroughly using the freeze-pump-thaw method. Then
the catalyst (∼9.1 × 10-5 mol) in CH2Cl2 (∼10 mL) was added at RT
under an inert atmosphere. The reaction was monitored by withdrawing
GC-samples at regular time intervals.
Membrane Experiments. All experiments were carried out at 25
°C in a 3.75 mL or a 10 mL polypropylene membrane reactor under
an inert Argon atmosphere (Argonbox) using the standard catalytic
JA0026612
(38) Spek, A. L. PLATON, a multipurpose crystallographic tool, Utrecht
University, The Netherlands, 1998.
(39) Beurskens, P. T.; Admiraal, G.; Beurskens, G.; Bosman, W. P.;
Garcia-Granda, S.; Gould, R. O.; Smits, J. M. M.; Smykalla, C. The
DIRDIF97 program system, technical report of the crystallographic labora-
tory; University of Nijmegen, The Netherlands, 1997.
(40) Sheldrick, G. M. SHELXL97, program for crystal structure refine-
ment; University of Go¨ttingen, Germany, 1997.