Proton-abstraction mechanism in the palladium-catalyzed intramolecular arylation: Substituent effects

Article abstract of DOI:10.1021/ja071034a

The regioselectivity observed in the intramolecular palladium-catalyzed arylation of substituted bromobenzyldiarylmethanes as well as theoretical results demonstrate that the Pd-catalyzed arylation proceeds by a mechanism involving a proton abstraction by

Full text of DOI:10.1021/ja071034a

Published on Web 04/27/2007
Proton-Abstraction Mechanism in the Palladium-Catalyzed
Intramolecular Arylation: Substituent Effects
Domingo Garc´ıa-Cuadrado, Paula de Mendoza, Ataualpa A. C. Braga,
Feliu Maseras,*,† and Antonio M. Echavarren*,‡
Contribution from the Institute of Chemical Research of Catalonia (ICIQ),
AV. Pa¨ısos Catalans 16, 43007 Tarragona, Spain
Received February 13, 2007; E-mail: fmaseras@iciq.es; aechavarren@iciq.es
Abstract: The regioselectivity observed in the intramolecular palladium-catalyzed arylation of substituted
bromobenzyldiarylmethanes as well as theoretical results demonstrate that the Pd-catalyzed arylation
proceeds by a mechanism involving a proton abstraction by the carbonate, or a related basic ligand. The
reaction is facilitated by electron-withdrawing substituents on the aromatic ring, which is inconsistent with
an electrophilic aromatic-substitution mechanism. The more important directing effect is exerted by electron-
withdrawing substituents ortho to the reacting site.
Scheme 1
Introduction
The arylation of arenes catalyzed by palladium is a more
direct alternative for the construction of biaryls1 than methods
based on cross-coupling reactions.2 Particularly useful has been
the intramolecular version that allows the cyclization of
substrates 1 to form carbo- and heterocycles 2 (Scheme 1).1,3,4
Intramolecular arylations are also involved in reactions mediated
by palladacycles.5-7
† Additional affiliation: Unitat de Qu´ımica F´ısica, Universitat Auto`noma
de Barcelona, 08193 Bellaterra, Catalonia, Spain.
‡ Additional affiliation: Departamento de Qu´ımica Orga´nica, Universidad
Auto`noma de Madrid, Cantoblanco, 28049 Madrid, Spain.
(1) (a) Hassan, J.; Se´vignon, M.; Gozzi, C.; Schulz, E.; Lemaire, M. Chem.
ReV. 2002, 102, 1359-1470. (b) Bringmann, G.; Tasler, S.; Pfeifer, R.-
M.; Breuning, M. J. Organomet. Chem. 2002, 661, 49-65. (c) Echavarren,
A. M.; Go´mez-Lor, B.; Gonza´lez, J. J.; de Frutos, OÄ . Synlett 2003, 585-
597. (d) Campeau, L.-C.; Fagnou, K. Chem. Commun. 2005, 1253-1264.
(e) Satoh, T.; Miura, M. J. Synth. Org. Chem. Jpn. 2006, 64, 1199-1207.
(f) Alberico, D.; Scott, M. E.; Lautens, M. Chem. ReV. 2007, 107, 174-
238.
(2) (a) de Meijere, A., Diederich, F., Eds. Metal-Catalyzed Cross-Coupling
Reactions; Wiley: Weinheim, Germany, 2004. (b) Espinet, P.; Echavarren,
A. M. Angew. Chem., Int. Ed. 2004, 43, 4704-4734. (c) Braga, A. A. C.;
Morgon, N. H.; Ujaque, G.; Maseras, F. J. Am. Chem. Soc. 2005, 127,
9298-9307.
(3) Lead references on synthetic applications: (a) Torres, J. C.; Pinto, A. C.;
Garden, S. J. Tetrahedron 2004, 60, 9889-9900. (b) Harrowven, D. C.;
Woodcock, T.; Howes, P. D. Angew. Chem., Int. Ed. 2005, 44, 3899-
3901. (c) Shen, D. M.; Liu, C.; Chen, Q. Y. J. Org. Chem. 2006, 71, 6508-
6511.
(4) (a) Campeau, L.-C.; Parisien, M.; Leblanc, M.; Fagnou, K. J. Am. Chem.
Soc. 2004, 126, 9186-9187. (b) Lafrance, M.; Blaquie´re, N.; Fagnou, K.
Chem. Commun. 2004, 2874-2875. (c) Parisien, M.; Valette, D.; Fagnou,
K. J. Org. Chem. 2005, 70, 7578-7584. (d) Campeau, L.-C.; Thansandote,
P.; Fagnou, K. Org. Lett. 2005, 7, 1857-1860. (e) Leblanc, M.; Fagnou,
K. Org. Lett. 2005, 7, 2849-2852. (f) Leclerc, J.-P.; Andre, M.; Fagnou,
K. J. Org. Chem. 2006, 71, 1711-1714. (g) Campeau, L.-C.; Parisien,
M.; Jean, A.; Fagnou, K. J. Am. Chem. Soc. 2006, 128, 581-590. (h)
Lafrance, M.; Blaquiere, N.; Fagnou, K. Eur. J. Org. Chem. 2007, 5, 811-
825.
In contrast to cross-coupling reactions,2 less has been known
on the mechanism of the Pd-catalyzed arylation. The initially
formed oxidative addition complex 3 could evolve by different
mechanisms via intermediates 4-6. An insertion into the arene
(Heck-like) would give 4,8 which should undergo a trans
â-hydrogen elimination to give 2. σ-Allyl palladium complex
4 might be in equilibrium with the corresponding η3-palladium
complexes. Most authors have favored an electrophilic aromatic
substitution (SEAr) via intermediates 5.9,10 However, in one case,
an intramolecular isotope effect kH/kD ) 3.5 has been
determined.4a A similar intramolecular isotope effect (kH/kD )
4) was found in the Pd-catalyzed synthesis of oxindoles that
proceeds via C-H functionalization.11 In addition, we have
(7) Ca´rdenas, D. J.; Mart´ın-Matute, B.; Echavarren, A. M. J. Am. Chem. Soc.
2006, 128, 5033-5040.
(8) Recent work indicates that this process is rather unlikely: Hughes, C. C.;
Trauner, D. Angew. Chem., Int. Ed. 2002, 41, 1569-1572.
(9) Substituent effects on the formation of five-membered-ring palladacycles
follows the order X ) MeO > H > NO2: (a) Catellani, M.; Chiusoli, G.
P. J. Organomet. Chem. 1992, 425, 151. (b) Mart´ın-Matute, B.; Mateo,
C.; Ca´rdenas, D. J.; Echavarren, A. M. Chem.sEur. J. 2001, 7, 2341-
2348.
(5) (a) Catellani, M.; Motti, E.; Faccini, F.; Ferraccioli, R. Pure Appl. Chem.
2005, 77, 1243-1248 and references therein. (b) Dupont, J.; Consorti, C.
S.; Spencer, J. Chem. ReV. 2005, 105, 2527-2571.
(6) (a) Zhao, J.; Campo, M.; Larock, R. C. Angew. Chem., Int. Ed. 2005, 44,
1873-1875 and references therein. See also: (b) Karig, G.; Moon, M.-T.;
Thasana, N.; Gallagher, T. Org. Lett. 2002, 4, 3115-3118. (c) Campo, M.
A.; Larock, R. C. J. Am. Chem. Soc. 2002, 124, 14326-14327.
(10) (a) Lane, B. S.; Sames, D. Org. Lett. 2004, 6, 2897-2900. (b) Lane, B. S.;
Brown, M. A.; Sames, D. J. Am. Chem. Soc. 2005, 127, 8050-8057. (b)
See, also: Park, C.-H.; Ryabova, V.; Seregin, I. V.; Sromek, A. W.;
Gevorgyan, V. Org. Lett. 2004, 6, 1159-1162.
(11) Hennessy, E. J.; Buchwald, S. L. J. Am. Chem. Soc. 2003, 125, 12084-
12085.
9
6880
J. AM. CHEM. SOC. 2007, 129, 6880-6886
10.1021/ja071034a CCC: $37.00 © 2007 American Chemical Society

Proton-Abstraction Mechanism
A R T I C L E S
Table 1. Pd-Catalyzed Arylation of Substrates 7a-la
found that the arylation on nitrofluorene12a or nitrocarbazole12b
gives substantial amounts of products by reaction at the positions
ortho or para to a strong electron-withdrawing NO2 group. An
interesting mechanistic alternative is a σ bond metathesis via
intermediates 6,11a which seems more likely than processes
involving C-H oxidative addition.13
Recently we advanced that the palladium-catalyzed arylation
reaction is facilitated by electron-withdrawing substituents on
the aromatic ring, which is inconsistent with an electrophilic
aromatic-substitution mechanism, and have proposed a proton-
abstraction mechanism for this reaction.14 Our mechanistic
proposal is in line with recent results on the intra-4g and
intermolecular15,16 palladium arylation that proceeds more easily
with arenes or heteroarenes bearing strongly electron-withdraw-
ing substituents. A related mechanism, in which acetate acts as
the basic ligand, has been proposed for cyclometallation of
benzylic amines with Pd(OAc)2.17
8/9
Herein we report a more thorough study on the effect of
substituents that more clearly reveals the directing effect that
substituents ortho to the reacting site exert on the palladium-
catalyzed arylation reaction.
T
yield
(%)
(8/8
′
/9)
entry substrate
substituent(s)
Lb
solvent
(
°
C)
ratio
1
2
3
4
7a
7a
7a
7b
R2 ) OMe
10a DMA 135 90 1.1:1
10b DMA 135 93 1.1:1
10a DMF 100 91 1.1:1
10a DMA 135 75 1.4:1
R2 ) OMe
R2 ) OMe
R1 ) OMe
Results and Discussion
(1.1:0.3:1)
Effects of Substituents: Experimental Results. To deter-
mine the mechanism of the Pd-catalyzed arylation, we decided
to study more precisely the effect of substituents on substrates
7a-l18 with an alkyl tether CH(R) between the two aryls, which
should present minimum steric and/or electronic bias in this
reaction. As standard conditions, we adopted those developed
by Fagnou,4a using bulky phosphine 10a as the ligand for Pd.
Under these conditions, the cyclization furnished the corre-
sponding 9,10-dihydrophenanthrenes, which in occasions were
obtained along with small amounts of phenanthrenes. To
5
6
7
8
9
7c
7d
7d
7d
7d
7e
R2 ) CF3
R2 ) Cl
R2 ) Cl
R2 ) Cl
R2 ) Cl
R1 ) Cl
10a DMA 135 71 1.3:1
10a DMA 135 66 1.5:1
10b DMA 135 66 2.4:1
10a DMA 100 98c 2:1
10a DMF 100 84 2:1
10a DMA 135 54 1.9:1
10
(0.8:1.1:1)
11
12
13
14
7f
R1 ) R2 ) R3 ) F 10a DMA 135 82 25:1
7g
7h
7i
R2 ) F
10a DMA 135 72 1.6:1
10a DMA 135 76 19:1
10a DMA 135 62 8.2:1
R1 ) R3 ) F
R1 ) R2 ) F
(6.8:1.4:1)
15
16
17
7j
7k
7l
R2 ) t-Bu
10a DMA 135 68 1:1.5
10a DMA 135 74 1:1.3
1
facilitate determining the ratio of regioisomers by H NMR
R2 ) SiMe3
R2 ) F, R4 ) t-Bu 10a DMA 135 81 2.3:1
analysis, the crude mixtures were treated with DDQ to give
phenanthrenes 8a-l/9a-l (Table 1). The structures of phenan-
threnes 9a,b,d,e,g,j,k were further confirmed by their indepen-
dent synthesis by Suzuki couplings of 9-bromophenanthrene
with the corresponding arylboronic acids.18
a 5% mol Pd(OAc)2, 10 mol % L, and 3 equiv K2CO3 for 16 h (DMA)
or 24 h (DMF). b 10a ) 2-(diphenylphosphino)-2′-(N, N-dimethylamino-
)biphenyl; 10b ) 2-(dicyclohexylphosphino)-2′,4′,6′-(triisopropyl)biphenyl.
1
c Yield determined by H NMR.
Similar regioisomeric ratios (1.1-2.4:1), favoring reaction
at the substituted aryl ring, were obtained from substrates 7a-e
bearing groups that are electron-releasing (OMe) or electron-
withdrawing (CF3, Cl) in SEAr processes (Table 1, entries
1-10). In the case of 7a a similar result was obtained with
ligand 10b19 (Table 1, entries 1 and 2). However, 10b led to a
better selectivity with substrate 7d (Table 1, entries 6 and 7).
Cyclizations could be also carried out in DMA or DMF at
100 °C (Table 1, entries 3, 8, and 9). Remarkably, reaction of
7f occurred almost exclusively at the trifluorophenyl ring to
give 8f (Table 1, entry 11). A single fluorine substituent in 7g
exerts a moderate control on the regioselectivity (Table 1, entry
12), similar to that of methoxyl, trifloromethyl, or chloro
substituents para to the arylation site. However, substrate 7h,
with fluorine substituents ortho to the arylation site (Table 1,
entry 13), led to 8h with an excellent regioselectivity (8h/9h )
19:1). The regioselectivity was ca. 8:1 with substrate 7i (Table
1, entry 14), although in this case the yield was lower. Taking
into account the fact that the two reactive positions at the phenyl
ring are identical, the selectivity for the reaction ortho to a
fluorine compared to the unsusbtituted aryl is ca. 14:1. Interest-
ingly, the arylation of substrates 7j and 7k with t-Bu and SiMe3
substituents, respectively, took place on the phenyl ring, with
moderate selectivity (Table 1, entries 15 and 16). The arylation
of substrate 7l was expected to take place selectively at the aryl
ring substituted with a fluorine with a 8l/9l regioselectivity of
ca. 2.4:1, calculated from the effects of entries 12 and 15 (1.6:1
(12) (a) Gonza´lez, J. J.; Garc´ıa, N.; Go´mez-Lor, B.; Echavarren, A. M. J. Org.
Chem. 1997, 62, 1286-1291. (b) Go´mez-Lor, B.; Echavarren, A. M. Org.
Lett. 2004, 6, 2993-2996.
(13) Mota, A. J.; Dedieu, A.; Bour, C.; Suffert, J. J. Am. Chem. Soc. 2005, 127,
7171-7182.
(14) Garc´ıa-Cuadrado, D.; Braga, A. A. C.; Maseras, F.; Echavarren, A. M. J.
Am. Chem. Soc. 2006, 128, 1066-1067.
(15) (a) Lafrance, M.; Rowley, C. N.; Woo, T. K.; Fagnou, K. K. J. Am. Chem.
Soc. 2006, 128, 8754-8756. (b) Lafrance, M.; Fagnou, K. J. Am. Chem.
Soc. 2006, 128, 16496-16497. (c) Lafrance, M.; Shore, D.; Fagnou, K.
Org. Lett. 2006, 8, 5097-5100.
(16) (a) Campeau, L.-C.; Rousseaux, S.; Fagnou, K. J. Am. Chem. Soc. 2005,
127, 18020-18021. (b) Leclerc, J.-P.; Fagnou, K. Angew. Chem., Int. Ed.
2006, 45, 7781-7786.
(17) Davies, D. L.; Donald, S. M. A.; Macgregor, S. A. J. Am. Chem. Soc.
2005, 127, 13754-13755.
(18) See supporting information for experimental details and characterization
data.
(19) Huang, X.; Anderson, K. W.; Zim, D.; Jiang, L.; Klapars, A.; Buchwald,
S. L. J. Am. Chem. Soc. 2003, 125, 6653-6655.
9
J. AM. CHEM. SOC. VOL. 129, NO. 21, 2007 6881

A R T I C L E S
Garc´ıa-Cuadrado et al.
Table 2. Intramolecular Isotope Effect on the Pd-Catalyzed
Arylation of 7ma
Table 3. Effect of the Base on the Pd-Catalyzed Arylation of
7d,ga
entry
substrate
base
solvent
T (°
C)
yield (%)
8/9 ratio
entry
solvent
T (°
C)
yield (%)
8m/9m ratio
kH/kD
1
2
3
4
5
7d
7d
7d
7d
7g
7g
7g
7g
7g
7g
K2CO3
Et3N
DBU
K-t-BuO
K2CO3
KHCO3
K3PO4
K2CO3/t-BuCO2H DMA
Na2HPO4
Na2HPO4
DMF
DMF
DMF
DMF
DMA
DMA
DMA
100
100
100
100
135
135
135
100
135
135
84
2:1
1
2
DMF
DMA
100
135
92
82
0.15:1
0.2:1
6.7
5.0
a Reaction run using 5% mol Pd(OAc)2, 10 mol % 10a, and 3 equiv
K2CO3 for 16 h.
b
72
1.6:1
1.5:1
1.6:1
1.7:1
6
68
7
66c
Scheme 2
8d
9
(>95)e
DMA
DMA
10
a Reaction run using 5% mol Pd(OAc)2, 10 mol % 10a, and 3 equiv
base for 16 h. b (1-(4-Chlorophenyl)ethane-1,2-diyl)dibenzene was observed
in the crude reaction mixture. c Reaction time ) 24 h (ca. 90% conversion).
d Reaction carried out in the presence of t-BuCO2H 30 mol % for 4 h.
1
e Conversion determined by H NMR.
of this mechanism is that the formation of the metal-carbon
bond is concerted with the breaking of the carbon-hydrogen
bond, with the hydrogen being transferred to a basic center.
The existence of this type of step does not fully define the
mechanism, and the three variations presented in Scheme 3 can
be postulated. Starting from the species resulting of the oxidative
addition of the C-Br bond, the proton can be transferred either
to the bromide ligand (unassisted path) or to another base in
the media (assisted path), which in the computation study is
hydrogen carbonate, a base that also leads to arylation (Table
3, entry 6). The external base can either coordinate previously
to the metal, displacing bromide (intramolecular assisted path),
or not (intermolecular assisted path).
We carried out a preliminary set of B3LYP calculations with
basis set I (see computational details in the Supporting Informa-
tion) on the three mechanisms of Scheme 3 on two different
model systems: [Pd(PH3)Br(o-((CH2)2-C6H5)Ph)] and [Pd(PH3)-
Br(o-((CH2)2-C6H2F3)Ph)]; labeled as t1a and t1b, respectively
(the t label stands for theoretical). These systems aim to
reproduce the attack in either of the two rings of the system
related to substrate 7f, which gave experimentally the highest
selectivity, 25:1. For each of the three mechanisms, we
computed the key transition state, and the reactant and product
it connects. The resulting transition states t1aTS for the case
of the hydrogen-substituted ring are presented in Figure 1. The
three geometries fit well in the proton-abstraction view: the
hydrogen atom being transferred is close to halfway between
carbon and oxygen (or bromine), and the three atoms are nearly
in linear arrangement. The potential energy barriers correspond-
ing respectively to the t1a and t1b systems are 43.3 and 34.5
kcal/mol (unassisted mechanism), 23.5 and 13.2 kcal/mol
(intramolecular-assisted mechanism), and 17.4 and 14.4 kcal/
mol (intermolecular-assisted mechanism). In all cases the species
with fluorine substituents in the ring is favored, which repro-
duces the experimental observation and confirms the validity
and 1:1.5, respectively). In the event, reaction of 7l under the
standard conditions led to 8l/9l in a 2.3:1 ratio (Table 1, entry
17).
To determine the possible intramolecular isotope effects in
the palladium-catalyzed arylation reaction, we subjected sub-
strate 7m to the standard conditions. Thus, reaction of 7m gave
0.2:1 (135 °C) and 0.15:1 (100 °C) ratios of 8m and 9m (Table
2, entries 1 and 2), which correspond to intramolecular isotope
effects kH/kD equal to 5.0 (135 °C) and 6.7 (100 °C), respec-
tively.
These results suggested that arylation of 5H-indeno[1,2-b]-
pyridine derivative 1118 would proceed selectively para to the
pyridine ring (Scheme 2). Indeed, reaction of 11 proceeded
smoothly at 100 °C to afford 12 as the major regioisomer (12/
13 ) 2.1:1).
Effect of Base. Satisfactory results were obtained with
K2CO3 as the base, whereas Et3N or DBU led to unchanged
starting material (Table 3, entries 2 and 3). Reduction of 7d
was observed when KO-t-Bu was used as the base (Table 3,
entry 4). On the other hand, KHCO3 and K3PO4 could also
be used, with only small erosions in the overall yield (Table
3, entries 6 and 7). Addition of pivalic acid15b led to similar
results (Table 3, entry 8). However, Na2HPO4 and NaH2PO4
were not effective (Table 3, entries 9 and 10). It is interesting
that the regioselectivities are largely independent from the base
used.
Computational Results: Overall Mechanism.20 These
experimental data are inconsistent with an aromatic electrophilic-
substitution mechanism, that should drive the C-H function-
alization away from the ring with electron-withdrawing fluorine
substituents. The reaction features do not suit well either with
the other traditionally proposed mechanisms indicated in Scheme
1, Heck-like or σ bond metathesis. Instead, the reaction seems
better fitted to the mechanism we dubbed as proton-abstraction,14
shown in Scheme 3. Variations of this mechanism have been
also proposed by other authors.13,15a,17,21,22 The defining feature
(21) Biswas, B.; Sugimoto, M.; Sakaki, S. Organometallics 2000, 19, 3895-
3908.
(22) Pinto, A.; Neuville, L.; Retailleau, P.; Zhu, J. Org. Lett. 2006, 8, 4927-
(20) See Supporting Information for computational details.
4930.
9
6882 J. AM. CHEM. SOC. VOL. 129, NO. 21, 2007

Proton-Abstraction Mechanism
A R T I C L E S
Figure 1. B3LYP optimized structures of the transition state t1aTS for the three mechanisms considered: unassisted (left), intramolecular assisted (middle),
and intermolecular assisted (right). Selected bond distances are given in Å.
Scheme 3
of the proton-abstraction scheme. Introduction of free-energy
corrections computed in gas-phase did not alter the trends in
the comparison between t1a and t1b. The computed free-energy
barriers for the two systems were, respectively, 37.6 and 29.5
kcal/mol (unassisted mechanism), 19.8 and 10.9 kcal/mol
(intramolecular-assisted mechanism), and 15.0 and 13.1 kcal/
mol (intermolecular-assisted mechanism). The results are similar,
and the use of gas-phase entropic corrections for reactions taking
place in solution is arguable. Because of this, we will not
consider this type of corrections in the rest of the work.
solution is insufficient, the assisted mechanisms may become
inaccessible.
The comparison between the two base-assisted mechanisms
is even less conclusive. In our preliminary communication we
favored the assisted intramolecular pathway, but we revise the
topic here in more detail. The lowest barrier (17.4 vs 23.5 kcal/
mol) corresponds to the intermolecular mechanism for the t1a
system with hydrogen substituents and to the intramolecular
mechanism (13.2 vs 14.4 kcal/mol) for the t1b system with
fluorine substituents. Further alterations of the otherwise similar
energetics can arise from steps not computed here, namely, the
replacement of bromide by hydrogencarbonate in the intramo-
lecular mechanism, and the departure of bromide in the
intermolecular mechanism (see Scheme 3). The use of hydro-
gencarbonate as a base is a reasonable assumption, but the
dianionic carbonate could also play a role. Unfortunately, our
attempts to introduce it in the calculation led to computational
artifacts (see Supporting Information), and we stayed with the
hydrogencarbonate model for all our studies. On the other hand,
The barrier for the unassisted process is more than 20 kcal/
mol higher than for the assisted processes. According to these
calculations, the unassisted mechanism seems thus unlikely. The
experimentally observed dependence of the nature of the base
on the reactivity (Table 2) points also in this direction, although
the base can play other roles in the catalytic cycle. The low
efficiency of the unassisted mechanism we find for our particular
system is not necessarily general. In fact, Fagnou and co-
workers16b have pointed out that the low solubility of the base
may hinder its role in the reaction. If the amount of base in
9
J. AM. CHEM. SOC. VOL. 129, NO. 21, 2007 6883

A R T I C L E S
Garc´ıa-Cuadrado et al.
Table 4. Computed Energy Barriers (kcal/mol) and Selectivity
Ratios for the Base-Assisted Intermolecular Abstraction Reaction
with Different Substrates
computed
ratio
experimental
ratio
entry
system
substituent(s)
barrier
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
t2a
t2b
t2c
t2d
t2e
t2f
t2g
t2h
t2i
12.4
5.6
6.2
11.5
12.3
11.7
5.8
1:1
1:1
25:1 (7f)
R1 ) R2 )R3 ) F
R1 ) F
4400:1
2100:1
3.0:1
1.1:1
2.4:1
3500:1
3050:1
3900:1
6.4:1
13.3:1
3.4:1
52:1
9.2:1
2.4:1
4.4:1
1:67000
1:1.1
1:3.0
R2 ) F
1.6:1 (7g)
R3 ) F
R4 ) F
R1 ) R2 ) F
R1 ) R3 ) F
R1 ) R4 ) F
R2 ) R3 ) F
R2 ) R4 ) F
R3 ) R4 ) F
R1 ) Cl
6.8:1 (7i)
19.0:1 (7h)
5.9
5.7
t2j
t2k
t2l
10.9
10.3
11.4
9.3
10.6
11.7
11.2
21.4
12.5
13.3
13
1.4:1 (7i)
0.8:1(7e)
1.5:1 (7d)
t2m
t2n
t2o
t2p
t2q
t2r
t2s
t2t
R2 ) Cl
R3 ) Cl
R4 ) Cl
1.1:1 (7e)
1:1.3 (7k)
R1 ) SiMe3
R2 ) SiMe3
R3 ) SiMe3
R4 ) SiMe3
Figure 2.
1:2.1
potassium hydrogen carbonate has been shown experimentally
to be an efficient base in the reaction of substrate 7g (Table 3,
entry 6).
in the computed reaction paths are further proof of the rich
mechanistic complexity of these reactions.
For the sake of simplicity, we will in any case choose one of
the two base-assisted mechanisms for the calculations that
follow. This will be the intermolecular-assisted pathway based
on the smaller difference this mechanism presents between the
t1a and t1b systems. For the intermolecular mechanism, the
difference in potential energies is 3.0 kcal/mol (17.4 vs 14.4
kcal/mol) and this would imply a proportion of activation of
fluorinated vs nonfluorinated rings of 40:1 at 135 °C, reasonably
close to the experimental data of 25:1 for the 7f substrate. For
the intramolecular mechanism, the difference is 10.3 kcal/mol,
which would lead to a preference of more than 700000:1. An
additional experimental result in support of the intermolecular
mechanism is the observation by Fagnou and co-workers15 that
the nature of the aryl halide affects significantly the regiose-
lectivity. This seems to imply that the halide is present on the
catalyst at the time of arylation, which would be incompatible
with the intramolecular mechanism. A final practical issue
favoring the choice of the intermolecular mechanism is that it
does not require the initial replacement of Br- by the base. This
step would likely have a low barrier but is strictly not necessary
in the intermolecular pathway.
Computational Results: Effect of Substituents. We evalu-
ated the barrier of the intermolecular-assisted mechanism for
the variety of systems t2 indicated in Figure 2. The tested
substituents were the same as in experiment, to further check
the validity of our postulated computational mechanism. Cal-
culation also allowed the consideration of systems, for instance
the whole series of monofluorinated and difluorinated systems,
that are difficult to access from an experimental point of view.
In this way, it is easier to analyze the sensibility of the selectivity
on the substituent position. The computational model was more
sophisticated for this set of calculations (see Supporting
Information). The full experimental phosphines were considered
through ONIOM(B3LYP:UFF) calculations. Basis set II was
used for the QM region, and solvation effects were introduced
through PCM calculations.
The computed barriers are summarized in Table 4. The
computed value of selectivity for the parent systems t2b/t2a
with three fluorine substituents is 2100:1, far from the 40:1
reported above with the smaller model system t1b/t1a, and even
further from the experimental value of 25:1 found for substrate
7f. This proves how sensitive the selectivity is to the computed
barrier difference, which changes only from 3.0 to 6.2 kcal/
mol. The computed selectivities are in general larger than the
experimental values. In spite of the effort of considering the
full experimental phosphine and the introduction of solvent
effects, there are still a number of simplifications, mentioned
above, that can explain this lack of quantitative agreement.
Among them, we are focusing in one single mechanism (assisted
intramolecular), using 1,2-diarylethane as diarene, and hydrogen
carbonate as the base. The overall experimental trend in the
substituent effect on selectivity is however maintained. Electron-
acceptor fluorine and chlorine substituents favor in all cases
the reaction, while electron-donor trimethylsilyl substituents
drive the reaction toward the other ring. The effect of fluorine
is in all cases much larger than that of chlorine and trimethyl-
silyl, with the sole exception of the case of trimethylsilyl in
position R1 (t2q), which will be discussed below.
The choice of one mechanism is certainly a simplification,
because both mechanisms can be operative, and the preference
for one of them may depend on the substrate. However, the
goal here is not to discern between the two mechanisms, but to
use a reasonable computational description that is able to account
for the main experimental features.
It is worth noticing that these same three mechanisms were
computationally analyzed in a related system by Woo, Fagnou,
and co-workers in a recent work,15a where these mechanisms
were labeled C1, C2, and C3. Their conclusion was that both
the unassisted and the intramolecular-assisted pathways could
explain the experimental results and that the intermolecular
assisted pathway was not feasible. The discrepancy with our
results is due to differences in the system studied, which in their
case is an intermolecular arylation, with no preexistent chemical
connection between the rings being coupled. These discrepancies
9
6884 J. AM. CHEM. SOC. VOL. 129, NO. 21, 2007

Proton-Abstraction Mechanism
A R T I C L E S
Figure 3. B3LYP optimized structures of the transition states t2qTS (left) and t2rTS (right) for the intermolecular base-assisted proton abstraction in
systems with a trimethylsilyl substituent. Selected bond distances are given in Å.
Selectivity is very sensitive to the position of the substituent
in the aryl ring. The example of the monofluorinated systems
(Table 4, entries 3 to 6) is very clear. The effect is much larger
in position 1, ortho to the C-H bond being activated. The
presence of three fluorines decreases the barrier of the unsub-
stituted parent system from 12.4 to 5.6 kcal/mol, and the
presence of only one fluorine substituent in the ortho position
is sufficient to bring it down to 6.2 kcal/mol. The placement of
single fluorine on any of the other three positions has in the
contrary an effect always smaller than 1 kcal/mol, with barriers
of 11.5, 12.3, and 11.7 kcal/mol. It is worth noticing that in the
case of fluorine, the effect decreases with the distance of the
fluorine from the activated C-H. The barrier is 6.2 kcal/mol
for the ortho position 1, 11.5 and 11.7 kcal/mol for the meta
positions 2 and 5, and 12.3 kcal/mol for the para position 3.
This dependence of the effect on the distance is clearly
suggesting an inductive effect. The results on the difluorine
derivatives (entries 7 to 12) confirm these trends. The larger
selectivity is by far associated to the presence of a fluorine in
the ortho position 1 (entries 7 to 9).
The results for trimethylsilyl (entries 17 to 20) fit also within
this model. Trimethylsilyl has an opposite effect to fluorine, in
agreement with its weak electron-releasing nature, and its effect
on selectivity is smaller, because it is more similar in electronic
properties to hydrogen, which is the competing substituent in
the other ring. A single additional factor appears in the case of
trimethylsilyl, which is the steric effect. This is a bulky
substituent and has a steric pressure in the position R1 that
greatly increases the energy of the corresponding transition state
(21.4 kcal/mol vs 12.4 corresponding to the parent system with
hydrogen substituents). The presence of this steric strain can
be seen in Figure 3, where the transition states for two of the
trimethylsilyl systems are shown. The steric repulsion is also
probably affecting the relative order of the barriers associated
to the other positions, with reaction via t2rTS being the least
hindered (see Figure 3). In any case, it is worth noticing that it
is not a purely steric effect, because even in this position the
barrier is higher (by 0.1 kcal/mol) than for the case of the system
with hydrogens. This difference is small, but it already gives a
selectivity of 1:1.1, close to the experimental observation of
1:1.3.
The case of chlorine (entries 13 to 16) follows in general the
same trends as for fluorine, albeit with smaller selectivities. This
is what one would expect from an inductive effect, and this
aspect of the results will not be further analyzed. The case of
entry 13, where the computed ratio (52:1) appears to be opposed
to the experimental value (0.8:1), deserves however some
discussion. This is the case corresponding to t2m, with a
chlorine in the ortho position. This ratio should be actually
corrected to 1.6:1, as there are two identical ortho positions at
the unsubstituted aryl and the 0.8 value corresponds to the
reaction at the position ortho to the chlorine substituent. Upon
analyzing the optimized structures, we found that the transition
states for most systems, and in particular t2m, t2n, t2o, and
t2p, are very similar between the real system, including the
bulky experimental phosphine, and the model system with PH3.
In contrast, this is not the case for the intermediate. Specifically,
t2m has a different conformational arrangement with the PH3
ligand, which is used for the calculation of the solvent effects.
We suspect that this is the origin of this irregular behavior,
which in fact arises from a computational artifact. This could
be probably corrected by geometry optimizations in solution,
which we were unable to carry out.
Conclusion
The intramolecular palladium-catalyzed arylation has been
studied on a variety of bromobenzyldiarylmethane systems.
Substituents are introduced in one of the aryl rings, and this
allows the investigation of their effect on the arylation process.
Electron-withdrawing substituents, such as fluorine, chlorine,
and trifluoromethyl favor the reaction on the substituted ring.
Electronegative methoxy groups that are electron-releasing in
electrophilic aromatic substitutions, behave similarly to chlorine
substituents, which indicates that the main effect of substituents
in this reaction is inductive. Electron-releasing substituents, like
tert-butyl or trimethylsilyl, drive the reaction toward the
unsubstituted ring. In addition, a substantial deuterium isotope
effect was determined in this process. These results are
incompatible with the electrophilic aromatic-substitution mech-
anism that had been previously favored for this type of arylation
processes. There is also a strong dependence of selectivity on
the position of the substituent, with the ortho position, closer
to the C-H bond being cleaved, being by far the most sensitive,
further suggesting that the substituents act on the selectivity
9
J. AM. CHEM. SOC. VOL. 129, NO. 21, 2007 6885

A R T I C L E S
Garc´ıa-Cuadrado et al.
through inductive effects. The performance of DFT calculations
reproduces the experimental trends and leads to the proposal
of a reaction mechanism where the key step is the abstraction
of a proton from the aryl ring by carbonate, hydrogencarbonate,
or a related basic ligand present in the reaction media.
information, E. Gonza´lez-Cantalapiedra for the synthesis of 5H-
indeno[1,2-b]pyridine, Dr. Sergio Pascual for additional experi-
ments on the effect of base, and Johnson Matthey PLC for
palladium salts.
Supporting Information Available: Experimental details,
characterization data, computational details, Cartesian coordi-
nates, and absolute energies. This material is available free of
charge via the Internet at http://pubs.acs.org.
Acknowledgment. This work was supported by the MEC
(CTQ2004-02869 to A.M.E. and CTQ2005-09000-C02-02/BQU
to F.M. and Consolider Ingenio 2010, Grant CSD2006-0003),
Project PicoInside, and the ICIQ Foundation. We thank Prof.
Timothy Gallagher for helpful comments and exchange of
JA071034A
9
6886 J. AM. CHEM. SOC. VOL. 129, NO. 21, 2007