Aryl-aryl bond formation by the fluoride-free cross-coupling of aryldisiloxanes with aryl bromides

Article abstract of DOI:10.1002/chem.201102285

The prevalence of the biaryl structural motif in biologically interesting and synthetically important molecules has inspired considerable interest in the development of methods for aryl-aryl bond formation. Herein we describe a novel strategy for this process involving the fluoride-free, palladium-catalysed cross-coupling of readily accessible aryldisiloxanes and aryl bromides. Using a statistical-based optimisation process, preparatively useful reaction conditions were formulated to allow the cross-coupling of a wide range of different substrates. This methodology represents an attractive, cost-efficient, flexible and robust alternative to the traditional transition-metal-catalysed routes typically used to generate molecules containing the privileged biaryl scaffold.

Full text of DOI:10.1002/chem.201102285

DOI: 10.1002/chem.201102285
Aryl–Aryl Bond Formation by the Fluoride-Free Cross-Coupling of
Aryldisiloxanes with Aryl Bromides
Christine M. Boehner,[a] Elizabeth C. Frye,[a] Kieron M. G. OꢀConnell,[a]
Warren R. J. D. Galloway,[a] Hannah F. Sore,[a] Patricia Garcia Dominguez,[b]
David Norton,[c] David G. Hulcoop,[c] Martin Owen,[c] Gillian Turner,[c]
Claire Crawford,[c] Helen Horsley,[d] and David R. Spring*[a]
Abstract: The prevalence of the biaryl
structural motif in biologically interest-
ing and synthetically important mole-
cules has inspired considerable interest
in the development of methods for
aryl–aryl bond formation. Herein we
describe a novel strategy for this pro-
cess involving the fluoride-free, palladi-
um-catalysed cross-coupling of readily
accessible aryldisiloxanes and aryl bro-
mides. Using a statistical-based optimi-
sation process, preparatively useful re-
action conditions were formulated to
allow the cross-coupling of a wide
range of different substrates. This
methodology represents an attractive,
cost-efficient, flexible and robust alter-
native to the traditional transition-
metal-catalysed routes typically used to
generate molecules containing the priv-
ileged biaryl scaffold.
À
Keywords: biaryls · C C coupling ·
cross coupling · silicon · synthetic
methods
Introduction
The importance of the biaryl structural motif in chemistry is
exemplified by its presence in a large number of biologically
active and functional molecules, including natural products,
pharmaceutical agents, catalysts and ligands (Scheme 1).[1,2]
Consequently the development of methods for aryl–aryl
bond construction has attracted immense interest amongst
synthetic chemists.[1–3] Transition-metal-catalysed cross-cou-
pling between a organometallic compound and an organic
halide or pseudo-halide represents the most widely utilised
strategy for biaryl bond formation.[1,2] Numerous organome-
tallic species have been utilized as suitable cross-coupling
partners, with organotin (Stille) and organoboron reagents
(Suzuki) being the most significant in terms of synthetic util-
ity.[1]
Scheme 1. Some examples of biologically interesting molecules that con-
tain biaryl systems. Bifonazole and boscalid have broad-spectrum activity
against fungal strains,[4,5] gilvocarcin V is known to have antitumour ac-
tivity[6] and fluorenone-based natural and synthetic compounds such as
dengibsinin are of pharmaceutical interest.[7] Biaryl phosphane ligands
such as 1 are widely employed in palladium-catalyzed amination reac-
tions.[8] In all cases, the biaryl bond is highlighted in bold
[a] Dr. C. M. Boehner, E. C. Frye, K. M. G. OꢀConnell,
Dr. W. R. J. D. Galloway, Dr. H. F. Sore, Dr. D. R. Spring
Department of Chemistry,University of Cambridge
Lensfield Road, Cambridge (UK)
Fax : (+44)1223-336362
E-mail: spring@ch.cam.ac.uk
Despite the undoubted value of such “traditional” cross-
coupling protocols in organic synthesis, there are drawbacks
associated with them; for example, the toxicity of organo-
stannanes and their by-products and difficulties associated
with preparation and purification of some organoboron re-
agents.[9–11] Consequently, there is continuing interest in the
development of novel coupling procedures for the formation
of biaryl bonds (and carbon–carbon bonds in general) that
utilise alternative organometallic coupling species and con-
[b] Dr. P. G. Dominguez
Departamento de Quimica, Universidade de Vigo (Spain)
[c] Dr. D. Norton, Dr. D. G. Hulcoop, Dr. M. Owen, Dr. G. Turner,
Dr. C. Crawford
GlaxoSmithKline, Gunnels Wood Road, Stevenage (UK)
[d] Dr. H. Horsley
UCB Pharma Ltd, 208 Bath Road, Slough (UK)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201102285.
13230
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 13230 – 13239

FULL PAPER
ditions. In this context, the use of organosilicon reagents in
palladium-catalysed cross-coupling has attracted the atten-
tion of synthetic chemists.[12–16] Organosilicon reagents have
a number of unique advantages over other organometallic
species including low toxicity, high stability, low molecular
weight and ease of introduction into various sub-
strates.[12,13,15]
tion.[16] The majority of work on masked silanol
ACHTNGUTERN(UNG ate)s has fo-
cused upon vinylation, either under fluoride-activated[33–40]
or fluoride free conditions;[30,41] though there are some ex-
amples of biaryl synthesis using masked silanol
ACHTUNGTRENUN(NG ate)s, most
require fluoride activation[42,43]
A continuing area of interest within our research group is
the use of disiloxanes as masked silanols in cross-coupling
reactions.[31,32] Disiloxanes have been shown to exist in equi-
librium with the corresponding silanolate species under
basic conditions (Scheme 2).[18,30,31] We have previously ex-
Seminal studies by Hiyama and Hatanaka established that
a range of alkenyl-, alkynyl and allyltrimethylsilanes could
undergo palladium-catalysed cross-coupling with aryl, vinyl
and allyl halides through activation of the silicon–carbon
bond with fluoride ions to facilitate transmetallation.[17]
Since these pioneering works, significant advances in the
cross-coupling of organosilicon reagents have been made
that have extended the scope and generality of this method
greatly. In the context of biaryl synthesis, numerous organo-
silane species have been successfully employed as aryl
donors in fluoride-activated cross-coupling reactions with
aryl halides.[18,19] However, there are several significant
drawbacks associated with the use of fluoride including the
cost of fluoride sources, the corrosive effects of such com-
pounds and its incompatibility with many functional groups,
particularly silicon protecting groups, which are ubiquitously
employed in organic synthesis.[11,12,18] Consequently, the de-
velopment of protocols that allow efficient coupling of aryl
silicon species with aryl halides under fluoride-free condi-
tions has attracted significant interest. For example, a varie-
ty of bases have been shown to act as effective activators of
silicon-based reagents.[20–28] In 2000, Hiyama and co-workers
described the fluoride-free palladium-catalysed cross-cou-
pling of organosilanol species with a variety of iodoarenes in
the presence of stoichiometric Ag2O as an activator.[29] Over
the last decade, the field of organosilanol-based cross-cou-
pling has developed rapidly, with significant practical and
theoretical contributions being made by Denmark and co-
workers.[11,12,14,18,30] These researchers have reported that ar-
omatic and heteroaromatic silanol species and pre-formed
silanolate salts can be cross-coupled with a range of aromat-
ic and heteroaromatic halides to form biaryl derivatives; the
reactive species in such processes has been shown to be the
corresponding silanolate, which in the case of silanols, is
generated in situ in the presence of a base.[11] Despite the
undoubted utility of these protocols for the formation of
biaryl bonds, the high levels of reactivity exhibited by
Scheme 2. Palladium-catalyzed, base-induced fluoride-free cross-coupling
of disiloxanes. [PdACTHNUTRGNEUNG(dba)2]=bis(dibenzylideneacetone)palladium(0).
ploited this phenomenon for the development of operation-
ally simple protocols for the base-induced cross-coupling of
a range of aryl substituted vinyldisloxanes 2 with aryl and
heteroaryl iodides and bromides under fluoride-free condi-
tions, providing access to (E)-stilbene derivatives 3 in good
to excellent yields with excellent levels of geometric purity
(Scheme 2).[31,32] Disiloxanes offer the advantage of in-
creased stability relative to silanolACTHNUTRGNEUNG(ate)s and therefore have
the potential to be carried through multi-step synthesis,
which could offer new disconnection strategies in the syn-
thesis of complex molecular systems.[31,32] In addition, the
use of disiloxanes is more atom efficient compared to alter-
native “masked” silanols, which have been more commonly
employed.[31,32]
silanolACHTUNGTRENNUNG(ate)s can hinder their progression through multistep
syntheses.[31,32] Consequently, complex synthetic sequences
incorporating silanol-based cross-coupling reactions are in-
herently inflexible in the sense that the relevant silcon func-
tionality cannot be installed at an early stage in the synthesis
due to an incompatibility with other functionality or re-
agents. In an effort to address this issue, various “masked”
We envisaged expanding the scope of our disiloxane-
based coupling strategy to include aryldisiloxanes 4, with
the aim of developing a novel, robust methodology of broad
synthetic utility and generality for the preparation of substi-
tuted biaryl systems 5. Herein we report upon the successful
realisation of this goal. Using statistical-based methods to
expedite the optimisation process, preparatively useful reac-
tion conditions were formulated that allow for the base-
mediated, palladium-catalysed, fluoride-free cross-coupling
of a wide range of readily prepared aryldisiloxanes 4 and
commercially available aryl bromides. These methods repre-
sent attractive, cost-efficient alternatives to the transition-
forms of alkenyl- and aryl silanolACTHUNTGRNEUNG(ate)s have been developed
and utilised to forge new carbon-carbon bonds.[16,32] Such
species are (ideally) stable to a wide range of reaction con-
ditions but can be selectively unmasked or “activated” in
situ under specific conditions to reveal the desired reactivate
silanol in preparation for the subsequent coupling reac-
Chem. Eur. J. 2011, 17, 13230 – 13239
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
13231

D. R. Spring et al.
metal-catalysed routes typically used to generate molecules
containing the privileged biaryl scaffold.
anes with aryl halides (data not shown). The cross-coupling
of arylACTHNURTGNEU(GN dimethyl)silanols with aryl halides has been report-
ed.[18] Inspired by these observations, we examined the use
of related reaction conditions for the cross-coupling of aryl-
disiloxanes (Table 2). However, the results of this initial
Results and Discussion
Preparation of disiloxanes: To comprehensively explore
aryl–aryl coupling employing aryldisiloxanes, a range of
these species, varying in both steric and electronic proper-
ties, were required. Towards this end, aryldisiloxanes 6–13
were synthesised from the appropriately substituted aryl
halide derivatives through halogen–lithium exchange and
trapping with chlorodisiloxane 14 (Table 1). Pleasingly, both
electron-rich and electron-poor aryl bromides readily par-
ticipated in the reaction and yields of the desired products
were typically good to excellent.
Table 2. Base-induced cross-coupling of aryldisiloxanes.[a]
Entry
1
Disiloxane
Product
Yield[b] [%]
NA
12
16
17
18
2
3
15[e]
15[e]
detected[c]
27[d]
Table 1. Synthesis of aryldisiloxanes using aryl lithium reagents.[a]
4
15[e]
19
detected[c]
5
6
7
9
8
6
20
21
22
28
À
Ar X
Entry
Product
Yield
[%][b]
detected[c]
11
1
2
6
7
96
94
[a] General conditions: disiloxane (1.0 equiv), aryl bromide (1.5 equiv).
[b] Yield of product isolated after purification by flash column chroma-
tography. Yield based upon aryl bromide. [c] Product mass was detected
in LCMS spectrum of the crude reaction mixture but could not be isolat-
ed. [d] Product yield determined by 1H NMR spectroscopic analysis of
the isolated mixture of 17 and the corresponding homocoupled product
after flash column chromatography. NA=no evidence for formation of
the desired product by LCMS and 1H NMR spectroscopic analysis of the
3
4
8
9
75
96
crude
reaction
mixture.
[e] 15=1,1,3,3-tetramethyl-1,3-bis-
(phenyl)disiloxane (R1 =Ph; commercially available).
5
10
81
study were disappointing; in several cases it did not prove
possible to isolate the desired biaryl product, and, in the
limited number of examples in which this could be achieved,
the isolated yields were typically low; homocoupled aryl
bromide and dehalogenated material were also observed as
significant by-products. Significant optimisation of the reac-
tion conditions was required to realise our goal of develop-
ing a synthetically useful protocol of broad generality for
the cross-coupling of aryldisiloxanes with aromatic halides.
Reaction optimisation is traditionally carried out by a
“one-factor-at-a-time-approach”, in which the effect of one
particular experimental variable (factor) is assessed by keep-
ing all other conditions constant, then this factor is held at
its optimum setting and a different factor is examined, and
so on.[44] The preliminary study outlined above suggested
that cross-coupling reactions involving aryldisiloxanes are
affected by many variables; thus a traditional systematic op-
timisation process would be expected to be an extremely
lengthy endeavour.[44] In addition, optimisation using a tradi-
tional approach is known to provide only a partial explora-
tion of the overall “reaction space” (the full range of all the
6
7
8
11
12
13
65
76
82
[a] General conditions: aryl halide (2.1 equiv), nBuLi (2.0 equiv) disilox-
ane (1.0 equiv). [b] Yield of product isolated after purification by flash
column chromatography.
Preliminary studies on biaryl formation using aryldisilox-
anes: With a range of aryldisloxanes in hand, their behav-
iour in palladium-catalysed base-mediated cross-coupling re-
actions with aryl bromides was investigated. Initially, the use
of reaction conditions we had previously developed for the
cross-coupling of vinyldisiloxanes with aryl halides were in-
vestigated; bis(dibenzylideneacetone)palladium(0) ([Pd-
ACHTUNGTRENNUNG
13232
www.chemeurj.org
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 13230 – 13239

Aryl–Aryl Bond Formation
FULL PAPER
variables and all possible combinations of the variables) as
any interactions between the variables are not considered;
thus a true reaction optimum may not be found.[44,45] We
therefore sought to optimise the reaction conditions for the
cross-coupling of aryldisiloxanes by using an alternative
strategy based on a statistical-based process, centred around
the application of a Design of Experiment (DoE) protocol.
DoE is a statistical tool used for experimental design and
the optimisation of continuous variables (i.e., number of
equivalents of given reagent, concentration of a given re-
agent, temperature, reaction time).[45] In the context of reac-
tion optimisation, use of DoE gives more precise informa-
tion about the reaction process (including knowledge about
the interaction(s) between different reaction variables) and
a fuller examination of the reaction space from fewer ex-
periments than that achieved using a traditional one-factor-
at-a-time-approach.[44–46] DoE thus offers the opportunity to
achieve a better understanding and control of a given reac-
tion process.[44,45] DoE was developed specifically to look at
large problems with many variables and interactions.[46]
Given that cross-coupling using aryldisiloxanes appeared to
be a complex process, with several variables influencing the
outcome of the reaction (see above), we believed that opti-
misation using a DoE strategy would be advantageous.
space than in previous investigations and 2) arrive at the de-
sired conditions in a reduced number of experiments. Im-
proving the yield of the desired biaryl product 18 was an ob-
vious objective for the optimisation process. Related objec-
tives included increasing the consumption of aryl bromide
23 and reducing the amount of homocoupled material 24
generated during such reactions.
The variables in this initial screening design included the
choice of ligand, base, solvent and the presence/absence of
H2O.[48] The palladium source used in all cases was chosen
to be [allylPdCl]2 as Denmark and Ober had previously
demonstrated its utility in the cross-coupling of aryl silanol-
AHCTUNGTRENNUNG
(ate)s.[18] Overall eight solvents, eight bases, and 16 ligands
were included (see the Supporting Information). A complete
set of all variable combinations, incorporating the presence/
absence of water would lead to 2048 experiments. However,
by using the screening design the number of experiments
was reduced to 68 (including four control replicates), while
still retaining good information on the suitability of the vari-
ables in the cross-coupling process. The responses in the
first screening design were defined as the amount (in%
yield) of product 18, aryl bromide 23, homocoupled product
24 and dehalogenated starting material 25. These were de-
termined through HPLC analysis of the crude reaction mix-
tures. More detailed information on this screening experi-
ment is provided in the Supporting Information. Ultimately,
the reaction components (discrete variables) selected as the
starting point for use in the DoE were: KOH (base),
Determination of the discrete variables for the DoE study
using an experimental design method: Before the DoE anal-
ysis could be carried out a specific set of discrete variables
(i.e., the exact types of reaction components such as base,
solvent and ligand) needed to be defined. The preliminary
investigations into the cross-coupling of aryldisiloxanes had
provided reaction conditions with limited substrate scope
which only gave the desired biaryl products in low yields
(see above). These conditions were thus deemed to be un-
suitable for use as a starting point for the DoE study. To
find a refined set of reaction components the reaction be-
tween disiloxane 15 and 4-bromo-tert-butylbenzene (23) to
produce 18 was examined (Scheme 3; this substrate pair was
also used as the “model” reaction in the subsequent DoE
analysis[47]). The effects of various discrete variables (differ-
ent bases, solvents and ligands) upon the reaction profile
was investigated, with experiments based upon a balanced
sub-set of potential combinations forming a categorical
screening design rather than a traditional one-variable-at-a-
time approach to: 1) cover a larger area of the reaction
tBuOH (solvent), PACTHNUTRGNEUNG(tBu)3 (ligand) in the presence of H2O.
Optimisation of continuous variables using a DoE protocol:
The DoE analysis involved a series of experiments which
analysed the effects of the continuous variables associated
with the reaction process; that is, the ranges of the reaction
components already identified (e.g. number of equivalents
of a specified reagent) and other variables such as reaction
time, reaction temperature, and so on. The ranges (high and
low levels) of the reaction variables chosen as the starting
point for the DoE study were:[49]
Base:
Catalyst:
H2O:
KOH, 1–10 equivalents
[allylPdCl]2: 2–10 mol%
0–20 equivalents
Temperature: 70–908C
Time: 3–10 h
A key goal of this study was the development of a process
that was competitive with the existing cross-coupling proce-
dures. Thus, an operationally simple protocol with relatively
mild reaction conditions and a broad substrate scope which
proceeded efficiently at low catalyst loading was desired;
these criteria guided the choice of upper limits for base, re-
action temperature, time and catalyst loading.
Initially a two-level factorial design with additional centre
points was carried out.[50] The responses in this DoE analysis
were defined as the amount (in% yield) of product 18 and
Scheme 3. Model substrate pair used in studies to determine discrete var-
iables for the DoE analysis.
Chem. Eur. J. 2011, 17, 13230 – 13239
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
13233

D. R. Spring et al.
Table 3. Details of experiments performed in two-level factorial design.[a]
and catalyst and its effect upon desired product for-
mation could be represented in a 2D and 3D plots,
which clearly indicate that the use of high equiva-
lents of H2O, high levels of base and catalyst load-
ing should lead to a high product yield (Figure 1
and Figure 2). The effect of temperature and time
proved to be only marginal.
The data gathered from the experimental design
indicated that the most important factor affecting
the yield of the undesired homocoupled material 24
was the interdependence of base and catalyst. The
analysis suggested that low catalyst loading together
with low amount of base should lead to a reduced
yield of this undesired material. However, these
conditions would be expected to lead to an inferior
yield of the desired cross-coupled material 18. Thus
the aim of discovering reaction conditions which
would provide the desired product 18 in a high
yield whilst minimising the quantity of homocou-
pled material 24 formed had not been realised by
this two-level factorial design. Clearly, further anal-
ysis of the reaction process was required.
In all the interaction graphs for both 18 and 24,
in which the remaining factors (i.e., those not in-
volved in the interaction under analysis) were set to
their centre point values, the centre point design
points (corresponding to the actual performed ex-
periments) were elevated relative to the interaction
lines (i.e., corresponded to a higher yield of the de-
sired product 18). That is, the centre points gave
yields of 18 which were above the values predicted
Run[b]
KOH
[equiv]
H2O
[equiv]
Cat
[mol%][c]
T
A
t
Yield
[%][f]
G
R
G
[o C][d]
[h][e]
Pr (18)
HC (24)
1
2
3
4
5
6
7
8
1.0
5.5
5.5
10.0
1.0
10.0
1.0
10.0
1.0
10.0
10.0
1.0
0
2.0
6.0
6.0
10.0
2.0
10.0
2.0
10.0
2.0
10.0
2.0
10.0
10.0
2.0
70
80
80
70
90
90
70
90
90
70
70
90
70
90
70
90
90
70
10
6.5
6.5
3
10
3
3
10
3
10
10
10
10
10
3
22
70
71
69
13
22
9
71
25
25
16
16
16
3
4
10.0
10.0
20.0
20.0
0
20.0
20.0
0
0
20.0
0
20.0
0
0
12
12
10
3
14
1
13
6
13
3
3
3
2
1
2
3
3
9
10
11
12
13
14
15
16
17
18
1.0
10.0
10.0
10.0
1.0
2.0
2.0
10.0
10.0
1
20.0
20.0
0
3
3
3
16
12
23
1.0
[a] General reaction conditions: disiloxane 15 (1.0 equiv), 4-bromo-tert-butylbenzene
(23) (1.8 equiv), Pd/P(tBu)3 1:1. [b] run=run order (computationally randomised
order, in which experiments are performed). [c] Cat=catalyst ([allylPdCl]2). [d] T=
temperature [e] t=time; h=hours. [f] Pr=product 18, HC=homocoupled product 24.
Yields (based upon aryl bromide) determined through HPLC diode array (UV-trace)
analysis of the crude reaction mixtures relative to the IS (1,1’-biphenyl-4-methanol).
ACHTUNGTRENNUNG
homocoupled product 24. The screening experiment was set
up using a statistical software package[51] and consisted of a
set of 18 reactions (Table 3). The responses were determined
through HPLC analysis of the crude reaction (see the Sup-
porting Information).
The effect of the various factors, including interacting fac-
tors, upon the yields of the desired product 18 and homo-
coupled material 24 were represented in a graphical fashion
in the form of main factor or interaction graphs and addi-
tionally as 3D-graphs (see the Supporting Information).
Significant factors affecting the product formation were
found to be catalyst, base and H2O. In addition, the follow-
ing interaction effects were also found to be significant:
base–H2O, base–catalyst, catalyst–H2O and time–tempera-
ture. There was also a smaller, negative interaction between
catalyst and temperature. The identification of these interac-
tion effects demonstrates the superiority of the DoE com-
pared to a one-factor-at-a-time approach for reaction opti-
misation, which does not take into account any interdepend-
ency of factors.[50] Further analysis indicated that the best re-
sults in terms of yield of the desired product 18 should be
achieved when all of these three factors were at their high-
est factor level. For example, the interaction between base
Figure 1. 2D Contour graph of the interaction of base and catalyst: Base
is displayed on the X axis, catalyst on the Y axis. The product (18) forma-
tion for the experimental space is represented in graduated colour shad-
ing, from blue (low yield) to bright red (high yield) and the contour lines
displaying the product yield. Equivalents of H2O is set to high factor
level (20.0 equiv), and time and temperature are set at their centre point
values.
13234
www.chemeurj.org
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 13230 – 13239

Aryl–Aryl Bond Formation
FULL PAPER
Figure 2. 3D view of the interaction of base and catalyst: Base is dis-
played on the X axis, catalyst on the Y axis. The product (18) formation
for the experimental space is displayed on the Z axis; H2O equivalents
are set to high level (20.0 equiv), time and temperature are set at their
centre point values (6.5 h, 808C).
by the computational model. This phenomenon suggested
considerable “curvature” in the experimental space for the
formation of the desired cross-coupled product 18 and the
homocoupled material 24, which indicated that the linear as-
sumptions made in the two-level factorial model were not
descriptive of the real response surfaces.[50] To model the
real reaction system with a higher degree of accuracy, a
more detailed study involving the examination of nonlinear
terms was carried out with the design points (experiments)
chosen according to a central composite design (CCD).[50]
A CCD involves experiments at extra positions of the
design space relative to a two-level factorial design to inves-
tigate the area around the centre points in the response
area.[50] The variables examined in this second study were
base, catalyst and water; all at the initial levels used in the
two-level factorial study.[52] CCD screening was set up using
a statistical software package.[51] Overall the experimental
design consisted of a total of 20 reactions. The responses
chosen were the yield of the desired product 18 as well as
the yield of the undesired homocoupled material 24, which
were determined through HPLC analysis of the reaction
mixtures. The effects of base–catalyst, base–water and cata-
lyst–water interactions upon the product formation could be
visualised in form of 2D contour plots and 3D response sur-
face plots. These graphs showed a gradual increase in yield
of the desired product 18 with increasing amounts of the
three main factors (base, catalyst, H2O), with the optimum
within this selected experimental space positioned at the
highest explored level of all these factors (Figure 3). As sug-
gested by the two-level factorial design, the formation of ho-
mocoupled product 24 is favoured by high catalyst loading
and high concentration of base. Unfortunately, these are the
conditions suggested to be necessary for a high yield of the
desired product 18. Overall, a better knowledge of the ex-
perimental space around the centre point region was gained
by the non-linear response surface study, which showed a
gradual increase in the yield of the desired product 18 to-
Figure 3. Upper: 3D response surface plot with catalyst loading displayed
on the X-axis, equiv of base on the Y axis and yield of 18 (%) on the Z
axis; Middle: 3D response surface plot with equiv of base displayed on
the X axis, equivalents of H2O on the Y axis and yield of 18 (%) on the
Z axis; Bottom: 3D response surface plot with catalyst loading displayed
on the X axis, equiv of H2O on the Y axis and yield of 18 (%) on the Z
axis. Actual point(s) included in the design space (that is, points corre-
sponding to actual performed experiments) that deviated significantly
from the computationally derived model are indicated by solid point(s).
wards a small plateau at the end of the design space; that is,
at a high level of each factor (and the data indicated that
further improvements in the yield of 18 could be obtained
with increases in the value of the main factors, that is, base,
catalyst, H2O).[53]
Determination of a final set of reaction conditions using a
multiple response optimisation process: In an attempt to
Chem. Eur. J. 2011, 17, 13230 – 13239
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
13235

D. R. Spring et al.
Table 4. Investigation of five computationally generated reaction solu-
tions.[a]
reduce the amount of base (10 equivalents) and, perhaps
more significantly, the catalyst loading (10 mol%), a multi-
ple response optimisation was performed.[51] Values of the
reagents (in this case KOH, H2O and [allylPdCl]2) within
defined limits, which were predicted to fulfil desired out-
comes (in this case, a high yield of the cross-coupled product
18 together with a low yield of homocoupled product 24)
were suggested by the statistical software.[51] The computa-
tional analysis predicted that only reaction conditions with a
catalyst loading ꢀ7 mol% would furnish the desired prod-
uct 18 in greater than 70% yield. Five examples of these
computationally generated suggestions were selected and
the reactions performed (Table 4). In terms of maximising
both the yield of 18 and the ratio of 18:24 whilst reducing
the amount of catalyst and base used, the best conditions
employed seven equivalents of base and seven mol% of cat-
alyst (Table 4, entry 4). An additional increase in yield of
the desired product 18 was obtained when the amount of
aryl bromide (4-bromo-tert-butylbenzene) was increased to
2.5 equivalents (Table 4, entry 6). Thus the optimised set of
reaction components for the cross-coupling of 15 and 23
Entry
H2O
[equiv]
KOH
[equiv]
[allylPdCl]2
[mol%]
Ratio
Yield
[%][c]
G
G
N
P/HC[b]
1
2
3
4
19
20
20
20
20
20
7
9
10
7
8
8
8
9
7
8
7
1:0.15
1:0.17
1:0.21
1:0.15
1:0.15
1:0.16
52
48
43
50
47
59
5
6[d]
7
[a] General reaction conditions: disiloxane 14 (1 equiv), 4-bromo-tert-bu-
tylbenzene (23) (1.8 equiv),Pd/ligand 1:2. [b] P=desired product 18,
HC=homocoupled product 24; ratio determined by the analysis of
1H NMR spectra of a mixture of 18 and 24 obtained after flash column
chromatography. [c] Isolated product yield of 18 after flash column chro-
matography (based upon aryl bromide). [d] 2.5 equiv of 4-bromo-tert-bu-
tylbenzene (23) used, yield based upon disiloxane (assuming 1 equiv disi-
loxane gives 2 equiv active silanolate).
were determined to be: [allylPdCl]2 (7.0 mol%), PACTHUNRGTNEUNG(tBu)3
(28.0 mol%), KOH (7.0 equivalents), H2O (20.0 equiva-
lents) and tBuOH at 808C with a reaction time of 20 h.
Exploring the substrate scope of the cross-coupling of aryl-
disiloxanes: Gratifyingly, the DoE-optimised reaction condi-
tions were found to be applicable to the cross-coupling of a
wide range of substituted aryldisiloxanes and aryl bromides
(Table 5). These included compounds 16, 17 and 19, which
could not be previously accessed (Table 5, entries 2 5, 9, 11
and 12), highlighting the expanded substrate scope offered
by the newly developed protocols. In addition, biaryls 16
and 19 could be accessed by using both appropriate sub-
strate combinations (Table 5, entries 5 and 7, and 2 and 9,
respectively) in comparable yields, which demonstrates the
flexibility of this cross-coupling strategy and is encouraging
for its future application in complex molecule synthesis. As
anticipated, the most frequently observed side product was
homocoupled material resulting from the self-condensation
of the aryl bromide component. The new methodology also
proved suitable for the reaction of heteroaryldisiloxane 13
(Table 5, entry 8) as well as aryl chlorides (formation of 19,
Table 5, entry 9). This last result is particularly appealing,
because aryl chlorides are lower cost and more widely avail-
able than aryl bromides.[54] Overall, the substrate scope of-
fered by this methodology is large, allowing rapid access to
a variety of biaryl products.
tors. Thus a statistical-based optimisation approach was
adopted, which centre upon the use of a DoE method to
more fully examine the reaction space and expedite the de-
velopment of robust reaction conditions. By this process, the
factors most affecting the yield of the desired cross-coupled
product were identified as catalyst loading, the amount of
base and the concentration of H2O. Preparatively useful re-
action conditions were formulated that allow for the cross-
coupling of a wide range of readily accessible aryl bromides
and electronically different aryldisiloxanes. Preliminary re-
sults indicate that it should be possible to expand the scope
of this disiloxane-based cross-coupling strategy to encom-
pass the use of heteroaromatic and aryl chloride coupling
partners. Further studies towards this end are ongoing, to-
gether with investigations into the application of this meth-
odology in multi-step synthesis. Overall, the procedures de-
scribed herein offer novel, cost-effective, operationally
simple and robust alternatives for the preparation of a wide
range of substituted biaryl systems. Some notable features
of this cross-coupling strategy in comparison with more “tra-
ditional” approaches include ease of disiloxane synthesis
and purification (compared with the preparation of organo-
boranes), low toxicity of disiloxanes and their by-products
(compared with organostannanes), increased functional
group tolerance in comparison with silanolACHTNUTRGNEUNG(ate)s (that could
Conclusion
offer new disconnection strategies in the synthesis of com-
plex molecular systems) and increased atom efficiency com-
pared with more commonly employed “masked” silanol spe-
cies. As such we anticipate that this general cross-coupling
strategy could prove valuable in a wide synthetic context,
with potentially broad applications in both target- and diver-
sity-oriented synthesis.[55,56]
Herein we have described the development of methodology
for biaryl synthesis based upon the palladium-catalysed,
base-induced, fluoride-free cross-coupling of aryl bromides
with aryldisiloxanes. Preliminary studies demonstrated that
such reactions were dependent upon many interacting fac-
13236
www.chemeurj.org
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 13230 – 13239

Aryl–Aryl Bond Formation
FULL PAPER
Table 5. Cross-coupling of aryldisiloxanes.
1
1
À
Si
À
Ar X
À
Si
À
Ar X
Entry
1[a]
R
Product
Yield
[%][f]
Entry
11[b]
R
Product
Yield
[%][f]
12
15
33
19
41
1
17
31
68
74
2[a]
53 (57[c]
)
12[b]
11
6
3[d]
6
25
63
13[b]
32
48
4[a]
10
12
34
16
38
51
14[b]
11
8
33
34
63
70
5[a]
15[d]
6[a]
6
6
28
16
72
49
16[d]
7
7
35
36
88
69
7[a]
17[d]
8[a]
13
15
7
29
19
30
24
29
62
18[e]
19[d]
20[d]
7
6
8
37
38
39
63
50
52
9[a]
10[b]
[a] Reagents as indicated, disiloxane (1.0 equiv), aryl halide (2.5 equiv), [allylPdCl]2 (7.0 mol%),
PACTHUNGETRNNU(G tBu)3 (28.0 mol%), KOH (7.0 equiv), H2O
(20.0 equiv), tBuOH, 808C, 20 h. [b] [allylPdCl]2 (10.0 mol%), P(tBu)3 (40.0 mol%). [c] NEt3 additive. [d] disiloxane (1.0 equiv), aryl halide (1.5 equiv),
AHCTUNGTRENNUNG
[allylPdCl]2 (5.0 mol%), dppb (10.0 mol%), KOH (3.0 equiv), H2O (2 equiv). [e] [allylPdCl]2 (10.0 mol%), dppb (20.0 mol%). [f] Isolated yield of prod-
uct after purification by flash column chromatography. For conditions [a] and [b] yields are based upon disiloxane (assuming 1 equiv disiloxane gives
2 equiv active silanolate). For conditions [d] yields are based upon aryl bromide. dppb=1,4-bis(diphenylphosphino)butane.
Acknowledgements
Experimental Section
The authors thank GlaxoSmithKline and UCB, the EU, EPSRC,
BBSRC, MRC, Wellcome Trust, and Frances and Augustus Newman
Details regarding the statistical processes described in this report and
practical experimental procedures and characterisation data for com-
pounds synthesised is provided in the Supporting Information.
Foundation for funding.
Chem. Eur. J. 2011, 17, 13230 – 13239
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
13237

D. R. Spring et al.
[32] H. F. Sore, C. M. Boehner, L. Laraia, P. Logoteta, C. Prestinari, M.
Scott, K. Williams, W. R. J. D. Galloway, D. R. Spring, Org. Biomol.
Chem. 2011, 9, 504.
[33] S. E. Denmark, D. Wehrli, J. Y. Choi, Org. Lett. 2000, 2, 2491.
[34] S. E. Denmark, J. Y. Choi, J. Am. Chem. Soc. 1999, 121, 5821.
[35] J. C. Anderson, S. Anguille, R. Bailey, Chem. Commun. 2002, 2018.
[36] S. E. Denmark, J. H. Liu, J. Am. Chem. Soc. 2007, 129, 3737.
[37] B. M. Trost, M. R. Machacek, Z. T. Ball, Org. Lett. 2003, 5, 1895.
[38] K. Itami, T. Nokami, J. I. Yoshida, J. Am. Chem. Soc. 2001, 123,
5600.
[1] J. Hassan, M. Sevignon, C. Gozzi, E. Schulz, M. Lemaire, Chem.
Rev. 2002, 102, 1359.
[2] G. P. McGlacken, L. M. Bateman, Chem. Soc. Rev. 2009, 38, 2447.
[3] M. Sainsbury, Tetrahedron 1980, 36, 3327.
[4] H. F. Avenot, T. J. Michailides, Plant Dis. 2007, 91, 1345.
[5] M. Botta, F. Corelli, F. Gasparrini, F. Messina, C. Mugnaini, J. Org.
Chem. 2000, 65, 4736.
[6] R. M. Knobler, F. B. Radlwimmer, M. J. Lane, Nucleic Acids Res.
1992, 20, 4553.
[39] W. Prukała, Synlett 2008, 3026.
[7] G. A. Salman, M. Hussain, A. Villinger, P. Langer, Synlett 2010,
3031.
[8] D. S. Surry, S. L. Buchwald, Angew. Chem. 2008, 120, 6438; Angew.
Chem. Int. Ed. 2008, 47, 6338.
[9] S. Napier, S. M. Marcuccio, H. Tye, M. Whittaker, Tetrahedron Lett.
2008, 49, 6314.
[10] G. A. Molander, N. Ellis, Acc. Chem. Res. 2007, 40, 275.
[11] S. E. Denmark, R. C. Smith, W. T. T. Chang, J. M. Muhuhi, J. Am.
Chem. Soc. 2009, 131, 3104.
[12] S. E. Denmark, M. H. Ober, Aldrichimica Acta 2003, 36, 75.
[13] S. E. Denmark, J. H. C. Liu, Angew. Chem. 2010, 122, 3040; Angew.
Chem. Int. Ed. 2010, 49, 2978.
[14] S. E. Denmark, J. Org. Chem. 2009, 74, 2915.
[15] S. E. Denmark, R. F. Sweis, Chem. Pharm. Bull. 2002, 50, 1531.
[16] A. C. Spivey, C. J. G. Gripton, J. P. Hannah, Curr. Org. Synth. 2004,
1, 211.
[17] Y. Hatanaka, T. Hiyama, J. Org. Chem. 1988, 53, 918.
[18] For a useful discussion of the variety of organosilanes used in biaryl
synthesis and the disadvantages of fluoride activation, see: S. E.
Denmark, M. H. Ober, Adv. Synth. Catal. 2004, 346, 1703, and refer-
ences therein. The authors also report a fluoride-free, base-activated
[40] S. E. Denmark, Z. G. Wang, Org. Lett. 2001, 3, 1073.
[41] J. C. Anderson, R. H. Munday, J. Org. Chem. 2004, 69, 8971.
[42] S. Napier, S. M. Marcuccio, H. Tye, M. Whittaker, Tetrahedron Lett.
2008, 49, 3939.
[43] A notable exception are the aryl(2-(hydroxymethyl)phenyl)dime-
thylsilanes developed by Hiyama and co-workers, which could be
successfully reacted with a broad range of aryl iodides in the pres-
ence of potassium carbonate and copper iodide to furnish biaryl de-
rivatives without the need for fluoride activators. See: a) Y. Nakao,
A. K. Sahoo, A. Yada, J. S. Chen, T. Hiyama, Sci. Technol. Adv.
Mater. 2006, 7, 536; b) Y. Nakao, J. S. Chen, M. Tanaka, T. Hiyama,
J. Am. Chem. Soc. 2007, 129, 11694; c) Y. Nakao, H. Imanaka, A. K.
Sahoo, A. Yada, T. Hiyama, J. Am. Chem. Soc. 2005, 127, 6952;
d) Y. Nakao, A. K. Sahoo, H. Imanaka, A. Yada, T. Hiyama, Pure
Appl. Chem. 2006, 78, 435. However, the use of aryl bromides in
such reactions is much less common; the development of a protocol
that allows the coupling of a wide range of aryl bromides under
mild fluoride-free conditions with a masked silanolACTHNUTRGNE(NUG ate)would be of
value as bromo derivatives are more readily available and typically
easier to synthesise than their iodo counterparts: e) see Ref. [31];
f) see Ref. [11].
coupling protocol using arylACTHUNTGRNEUNG(dimethyl)silanols.
[19] For examples of organosilane species that have been successfully
employed as aryl donors in fluoride-activated cross-coupling reac-
[44] M. R. Owen, C. Luscombe, L. W. Lai, S. Godbert, D. L. Crookes, D.
Emiabata-Smith, Org. Process Res. Dev. 2001, 5, 308.
[45] H. Tye, Drug Discovery Today 2004, 9, 485.
[46] D. Lendrem, M. Owen, S. Godbert, Org. Process Res. Dev. 2001, 5,
324.
tions with aryl halides see: for arylACTHNUTRGNEUGN(fluro)silanes: a) Y. Hatanaka, T.
Hiyama, Tetrahedron Lett. 1990, 31, 2719; b) Y. Hatanaka, S. Fu-
kushima, T. Hiyama, Chem. Lett. 1989, 1711; c) Y. Hatanaka, K.
Goda, Y. Okahara, T. Hiyama, Tetrahedron 1994, 50, 8301; aryl-
[47] It was thought that to achieve a significant improvement of the
cross-coupling process (in terms of substrate scope and yield) chal-
lenging substrates with low reactivity should be chosen; the expecta-
tion was that conditions optimised for such a pair would translate to
a wider range of disiloxanes and aryl bromides than if optimisation
were carried out using more reactive substrates. We decided upon
disiloxane 15 and 4-bromo-tert-butylbenzene (23) as the substrate
pair; they furnished the corresponding biaryl product in a low yield
(27%) using the reaction conditions developed from the preliminary
studies (Table 2, entry 3) and furthermore both compounds are com-
mercially available, relatively inexpensive and easy to handle.
[48] It had previously been noted that the presence of H2O was advanta-
geous in reactions involving the cross-coupling of silanols, probably
due to a shifting of the aryldisiloxane equilibrium to towards the re-
active silanolate species (see Ref. [18]). However, in combination
with other solvents or bases water could have detrimental effects
upon the cross-coupling reaction. Therefore the presence of water
was included as one of the variables.
[49] In a DoE, the ranges of the factors to be investigated (defining the
covered experimental space) should be sufficiently wide so as to
ensure that the optimum is residing within the experimental space
studied, but not so large that experiments become impractical (e.g.,
an extremely long reaction time) or indeed, that the overall process
yields optimised values for a given reaction component that are un-
desirable (e.g., a very large catalyst loading). See Ref. [50].
[50] V. K. Aggarwal, A. C. Staubitz, M. Owen, Org. Process Res. Dev.
2006, 10, 64.
ACHTUNGTRENNUNG
ACHTUNGTRENNUNG
1, 1495; aryl (trialkoxy)silanes: g) M. E. Mowery, P. DeShong, J.
Org. Chem. 1999, 64, 1684; h) M. E. Mowery, P. DeShong, Org. Lett.
1999, 1, 2137; i) H. M. Lee, S. P. Nolan, Org. Lett. 2000, 2, 2053;
j) M. L. Clarke, Adv. Synth. Catal. 2005, 347, 303; aryl-
ACHTUNGTRENNUNG(triallyl)silanes: k) Y. Nakao, T. Oda, A. K. Sahoo, T. Hiyama, J. Or-
ganomet. Chem. 2003, 687, 570; l) A. K. Sahoo, Y. Nakao, T.
Hiyama, Chem. Lett. 2004, 33, 632.
[20] E. Hagiwara, K. Gouda, Y. Hatanaka, T. Hiyama, Tetrahedron Lett.
1997, 38, 439.
[21] M. Murata, R. Shimazaki, S. Watanabe, Y. Masuda, Synthesis 2001,
2231.
[22] S. N. Chen, W. Y. Wu, F. Y. Tsai, Tetrahedron 2008, 64, 8164.
[23] C. Wolf, R. Lerebours, Org. Lett. 2004, 6, 1147.
[24] B. C. Ranu, K. Chattopadhyay, L. Adak, A. Saha, S. Bhadra, R.
Dey, D. Saha, Pure Appl. Chem. 2009, 81, 2337.
[25] E. Alacid, C. Najera, Adv. Synth. Catal. 2006, 348, 945.
[26] M. Suguro, Y. Yamamura, T. Koike, A. Mori, React. Funct. Polym.
2007, 67, 1264.
[27] M. Endo, T. Sakurai, S. Ojima, T. Katayama, M. Unno, H. Matsu-
moto, S. Kowase, H. Sano, M. Kosugi, K. Fugami, Synlett 2007, 0749.
[28] T. S. Huang, C. J. Li, Tetrahedron Lett. 2002, 43, 403.
[29] E. Hirabayashi, A. Mori, J. Kawashima, M. Suguro, Y. Nishihara, T.
Hiyama, J. Org. Chem. 2000, 65, 5342.
[30] S. E. Denmark, C. R. Butler, J. Am. Chem. Soc. 2008, 130, 3690.
[31] H. F. Sore, C. M. Boehner, S. J. F. MacDonald, D. Norton, D. J.
Foxd, D. R. Spring, Org. Biomol. Chem. 2009, 7, 1068.
[51] DesignExpert 8, 45 day trial version, StatEase, Minneapolis, 2010.
[52] Although the data obtained from the two-level factorial study sug-
gested that values for catalyst loading, base concentration and
equivalents of H2O above the upper levels examined (i.e., outside
the design space explored) should lead to a higher yield of the de-
13238
www.chemeurj.org
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 13230 – 13239

Aryl–Aryl Bond Formation
FULL PAPER
sired product 18, we decided to keep the initial factor levels for the
CCD analysis, as an increased amount of these factors, particularly
base and catalyst, would limit the attractiveness of the methodology.
Rather, we hoped to discover an optimum within the constraints of
the design space, with base and catalyst levels lower than their high-
est levels.
material 24 cannot be minimised without negatively impacting upon
the yield of the desired product 18.
[54] M. L. Clarke, M. B. France, J. A. Fuentes, E. J. Milton, G. J. Roff,
Beilstein J. Org. Chem. 2007, 3, 18.
[55] W. R. J. D. Galloway, A. Isidro-Llobet, D. R. Spring, Nature
Commun. 2010, 1, 80.
[53] Such conditions also favour formation of the undesired homocou-
pled material 24. Indeed, from these analyses it appears that this
process cannot be “decoupled” from formation of the desired prod-
uct to any great extent; that is, the yield of undesired homocoupled
[56] W. R. J. D. Galloway, D. R. Spring, Nature 2011, 470, 43.
Received: July 25, 2011
Published online: October 11, 2011
Chem. Eur. J. 2011, 17, 13230 – 13239
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
13239