Ligand-Controlled Regiodivergent Pathways of Rhodium(III)-Catalyzed Dihydroisoquinolone Synthesis: Experimental and Computational Studies of Different Cyclopentadienyl Ligands

Article abstract of DOI:10.1002/chem.201404515

Rh^III-catalyzed directed C-H functionalizations of arylhydroxamates have become a valuable synthetic tool. To date, the regioselectivity of the insertion of the unsaturated acceptor into the common cyclometalated intermediate was dependent solely on intrinsic substrate control. Herein, we report two different catalytic systems that allow the selective formation of regioisomeric 3-aryl dihydroisoquinolones and previously inaccessible 4-aryl dihydroisoquinolones under full catalyst control. The differences in the catalysts are computationally examined using density functional theory and transition state theory of different possible pathways to elucidate key contributing factors leading to the regioisomeric products. The stabilities of the initially formed rhodium complex styrene adducts, as well as activation barrier differences for the migratory insertion, were identified as key contributing factors for the regiodivergent pathways. Rh^III-catalyzed directed C-H functionalization of aryl hydroxamates enables the selective formation of regioisomeric 3-aryl or 4-aryl dihydroisoquinolones with two different catalytic systems. The different selectivities are examined using DFT and transition state theory. The stabilities of the rhodium complex adducts and migratory-insertion activation barriers are key factors for the regiodivergent pathways (see scheme; rs=regioselectivity).

Full text of DOI:10.1002/chem.201404515

DOI: 10.1002/chem.201404515
Full Paper
&
CꢀH Functionalization |Hot Paper|
Ligand-Controlled Regiodivergent Pathways of Rhodium(III)-
Catalyzed Dihydroisoquinolone Synthesis: Experimental and
Computational Studies of Different Cyclopentadienyl Ligands
Matthew D. Wodrich,[a] Baihua Ye,[b] Jꢀrꢁme F. Gonthier,[a] Clꢀmence Corminboeuf,*[a] and
Nicolai Cramer*[b]
Abstract: RhIII-catalyzed directed CꢀH functionalizations of
arylhydroxamates have become a valuable synthetic tool. To
date, the regioselectivity of the insertion of the unsaturated
acceptor into the common cyclometalated intermediate was
dependent solely on intrinsic substrate control. Herein, we
report two different catalytic systems that allow the selective
formation of regioisomeric 3-aryl dihydroisoquinolones and
previously inaccessible 4-aryl dihydroisoquinolones under
full catalyst control. The differences in the catalysts are com-
putationally examined using density functional theory and
transition state theory of different possible pathways to elu-
cidate key contributing factors leading to the regioisomeric
products. The stabilities of the initially formed rhodium com-
plex styrene adducts, as well as activation barrier differences
for the migratory insertion, were identified as key contribu-
ting factors for the regiodivergent pathways.
Introduction
erocycles, often providing good substrate-controlled regiose-
lectivities. In this respect, the synthesis of dihydroisoquino-
lones from pivaloyl arylhydroxamates[8] and an olefin inde-
pendently reported by Glorius[7b] and Fagnou[7c] is an important
example. Terminal olefins with electron-withdrawing or func-
tional groups such as arenes and esters form relatively selec-
tive 3-substituted dihydroisoquinolones (Scheme 1). Simple
alkyl alkenes give no or only very low regioselectivities, while
Addressing the issues of chemo-, regio-, diastereo-, and enan-
tioselectivity is of utmost importance for synthetic chemistry[1]
with great effort being made to develop highly selective reac-
tions. Understanding the origins and mechanisms of these ex-
perimentally observed, yet often puzzling, selectivities is also
beneficial to enhance synthetic methodology. A great strength
of computational approaches, particularly electronic structure
theory, is the ability to unravel puzzles of this type through the
elucidation of detailed reaction mechanisms.[2] Over the past
decade, selective transition-metal catalyzed CꢀH bond func-
tionalizations have become a powerful synthetic strategy.[3] For
instance, CꢀH activations induced by rhodiumIII catalysts have
emerged as versatile tools for directed C(sp2)ꢀH functionaliza-
tions.[4] To date, most experimental focus has been placed on
increasing the range of suitable directing groups to induce the
CꢀH activation or expanding the acceptor substrate scope. A
wide range of acceptor electrophiles, including alkynes,[5] al-
lenes,[6] and olefins,[7] have been used for the synthesis of het-
Scheme 1. Substrate-controlled regioselectivity in dihydroisoquinolone for-
mation by RhIII-catalyzed CꢀH functionalization.
[a] Dr. M. D. Wodrich,+ Dr. J. F. Gonthier, Prof. Dr. C. Corminboeuf
Laboratory for Computational Molecular Design
many of the more challenging 1,2-disubstituted alkenes do not
react at all. The access to the opposite 4-substituted product
isomer is scarce and limited to two heteroatom-substituted
olefins. We found that vinyltrimethylsilane selectively gives the
4-substituted product[9] and Molander discovered that vinyltri-
fluoroborate also displays this reversed regioselectivity.[10] How-
ever, a direct and general access to the synthetically valuable
4-aryl substituted dihydroisoquinolones is highly desirable.
Besides the steric and electronic nature of the accepting
olefin, the substituent/leaving group of the hydroxamate
Institute of Chemical Sciences and Engineering
Ecole Polytechnique Fꢀdꢀrale de Lausanne (EPFL), Lausanne (Switzerland)
E-mail: clemence.corminboeuf@epfl.ch
[b] B. Ye,+ Prof. Dr. N. Cramer
Laboratory of Asymmetric Catalysis and Synthesis
Institute of Chemical Sciences and Engineering
Ecole Polytechnique Fꢀdꢀrale de Lausanne (EPFL), Lausanne (Switzerland)
E-mail: nicolai.cramer@epfl.ch
[+] These authors contributed equally to this work.
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201404515.
Chem. Eur. J. 2014, 20, 1 – 11
1
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
group is of importance to the reaction outcome.[9] With
a methyl group, b-H elimination is predominant, providing
Heck-type products, whereas pivaloyl groups lead to the de-
sired reductive CꢀN bond formation concomitantly occurring
with NꢀO bond cleavage to close the catalytic cycle. The
nature of this effect has been computationally detailed by
Xia.[11] We have previously found that a Boc group instead of
pivaloyl has a remarkable influence on the selectivity of the re-
action.[9]
been mechanistically[7c] and computationally[7c,11] studied, the
determining factors for the regioselective olefin insertion
remain unknown. In particular, how the orientation of an unsa-
turated acceptor to the metal complex dictates the regioselec-
tivity has not been investigated and remains poorly under-
stood. However, a comprehensive understanding at a high the-
oretical level is important and invaluable in designing better
and more regioselective catalysts. Furthermore, it provides
a solid basis on which to develop an understanding of the
enantioselective variants of this type of catalysis. Herein, we
report a catalyst-controlled regiodivergent access to dihydro-
quinolones coupled with computational analysis to better ra-
tionalize the experimentally established selectivities.
To overcome variable substrate-dictated selectivities of a par-
ticular substrate, a significant advance would be complete re-
giocontrol by the employed catalyst. To put this appealing
strategy into practice, we conceived the use of different types
of cyclopentadienyl ligands of the RhIII complex to switch se-
lectivities and to conduct the reactions under full catalytic con-
trol. However, few reports have used ligands other than Cp* to
improve reactivities or selectivities[12] and altering the sub- Results and Discussion
strate-directed regiochemical outcome has been largely ne-
Regioselectivity experimental investigations
glected.[13] To date, examples involving the switching of selec-
tivities by the nature of the Cp ligand have, to our knowledge,
been elusive for rhodiumIII catalysis. Chiral Cp ligands permit-
ting access to the biologically relevant dihydroisoquinolone
scaffold enantioselectively have been recently independently
reported by Rovis, Ward,[14] and us.[9] The success of this enan-
tioselective incorporation of olefins prompted us to investigate
the effect of different cyclopentadienyl ligands that might
allow switching of the incorporation regiochemistry. We were
particularly interested in the decisive factors playing a role in
the product distribution and how they could be manipulated.
The key intermediate is the fluxional rhodacycle A, which co-
ordinates with alkene 2 (Scheme 2). With a terminal olefin,
there are in total four possible coordination modes to A.[15]
Formation of the B1 and B2 species would lead to enantiomer-
ic 4-aryl dihydroisoquinolones 4, whereas species B3 and B4
would provide enantiomeric 3-aryl dihydroquinolones 3. While
the initial cyclometalation step has received much interest and
For an initial evaluation of the influence of different cyclopen-
tadienyl ligands on the regioselectivity, we selected the hy-
droxamate, 1a, and styrene, 2a, as model substrates. Ideally,
the selective formation of either 3-substituted dihydroisoqui-
nolone, 3aa, or 4-substituted dihydroisoquinolone, 4aa, could
be triggered with the proper choices of rhodiumIII catalyst and
reaction conditions (Table 1). In situ preparation of the rhodi-
umIII catalyst by rapid oxidation of the corresponding rhodiumI
precursor with (BzO)2 gave the same results as the direct usage
of preformed rhodiumIII benzoate species. The simple [Cp*Rh-
(OBz)2] complex provided, under mild standard reaction condi-
tions at ambient temperature in ethanol, good selectivity for
the unexpected regioisomer 4aa (88:12, 92% yield; Table 1,
entry 1). This regioselectivity has never been previously repor-
ted.[7b,c] The evaluation of other complexes revealed that Rh3,
based on Tanaka’s electron-poor Cp* analogue,[12a] or Rh4,
mimicking the electronics of an unsubstituted Cp ligand,[16]
were both rather inefficient for this transformation (Table 1, en-
tries 4 and 5). The tBu2Cp-bearing complex, Rh5, previously
used by Rovis,[12b,c] was also not efficient (Table 1, entry 6). In
stark contrast, a simple 1,2-disubstituted Cp-containing com-
plex, Rh1, exclusively produced the 3-isomer 3aa in good yield
(Table 1, entry 2). Within detection limits, no 4-isomer 4aa was
found. Although the selectivity of the simplified cyclohexyl ver-
sion, termed CpCy (Rh2; Table 1, entry 17), is lower than that
for Rh1, its use might be preferred over Rh1 due to its simple
2-step synthesis. High regioselectivity is observed in a range of
alcoholic solvents (Table 1, entries 7–9). Trifluoroethanol is
slightly better (90:10) than ethanol (88:12; Table 1, entries 1
and 9), whereas hexafluoroisopropanol (HFIP) leads to decom-
position of 1a (Table 1, entry 10). On the other hand, the selec-
tivity of the Cp*Rh-complex decreases drastically to 60:40 in
nonprotic or apolar solvents, such as toluene, CH2Cl2, acetone
and THF (Table 1, entries 11–14). Two sets of conditions
emerged to selectively address the synthesis of both regioiso-
meric products; using Rh2 in CH2Cl2 (conditions A) to access 3-
aryl dihydroisoquinolones and [Cp*Rh(OBz)2] in trifluoroethanol
(conditions B) for 4-aryl dihydroisoquinolones.
Scheme 2. Regioselectivity-determining steps for catalyst-controlled olefin
incorporation in cyclometalated intermediate A.
&
&
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
2
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
Table 1. Influence of the different reaction parameters on the regiodiver-
gent styrene incorporation.[a]
Table 2. Scope for regiodivergent synthesis of dihydroisoquinolones 3
and 4.[a]
Entry
1
Conditions A
Conditions B
Entry
Rh Catalyst
Solvent
Total Yield [%]
4aa/3aa
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
[Cp*Rh(OBz)2]
Rh1+(BzO)2
Rh2+(BzO)2
Rh3+(BzO)2
Rh4+(BzO)2
Rh5+(BzO)2
[Cp*Rh(OBz)2]
[Cp*Rh(OBz)2]
[Cp*Rh(OBz)2]
[Cp*Rh(OBz)2]
[Cp*Rh(OBz)2]
[Cp*Rh(OBz)2]
[Cp*Rh(OBz)2]
[Cp*Rh(OBz)2]
[Cp*Rh(OBz)2]
Rh2+(BzO)2
Rh2+(BzO)2
Rh2+(BzO)2
EtOH
CH2Cl2
EtOH
EtOH
EtOH
92
73
53
88:12
1:>99
4:96
3aa, 72% (94:6)
4aa, 80% (90:10)
<5 (30% conv.)
8 (45% conv.)
16 (40% conv.)
93
–
50:50
50:50
84:16
82:18
90:10
–
60:40
72:28
73:27
60:40
63:37
3:97
EtOH
2
3
4
MeOH
iPrOH
CF3CH2OH
HFIP
toluene
CH2Cl2
acetone
THF
CH3CN
toluene
CH2Cl2
CH3CN
96
91 (80% 4aa)[b]
0
94
97
93
99
80
61
3ab, 65% (94:6)
4ab, 74% (90:10)
76 (72% 3aa)[b]
12 (50% conv.)
6:94
27:73
3ac, 63% (93:7)
4ac, 72% (90:10)
[a] Conditions: 1a (0.10 mmol), 2a (0.20 mmol), Rh catalyst (5 mol%),
0.2m in the indicated solvent, 238C, 14 h; yields and ratios were deter-
mined by 1H NMR spectroscopy of the crude product with an internal
standard; [b] isolated product.
3ad, 68% (>99:1)
4ad, 75% (86:14)
The scope for the regiodivergent styrene incorporation was
subsequently investigated using the two aforementioned opti-
mized conditions. A range of sterically and electronically modi-
fied aryl hydroxamates 1 and styrenes 2 were evaluated
(Table 2). Using conditions A with the 3-aryl selective catalyst
Rh2, the regioselectivities and the yields of the products 3
compare well with the standard substrate combination. When
using aryl hydroxamates having an electron-withdrawing
group in their para position, such as 1e and 1 f, yields and re-
gioselectivities (92:8) were somewhat lower. However, using
Rh1 as catalyst for these cases, higher yields and largely in-
creased regioisomeric ratios (>99:1) were obtained. The same
set of substrates 1 and 2 were submitted to conditions B using
[Cp*Rh(OBz)2] position 4-selective catalyst. Similar to the other
set of conditions, the regioisomeric ratio is mostly constant
(86:14–91:9) independent of steric and electronic substrate
modifications. In all cases, the minor product 3 is removable
by chromatography and good yields of the pure isolated 4-aryl
dihydroisoquinolones 4 were obtained.
5
3ae, 63% (96:4)
4ae, 75% (86:14)
6
3af, 66% (94:6)
3ba, 72% (94:6)
3ca, 63% (94:6)
4af, 54% (88:12)
4ba, 53% (86:14)
4ca, 80% (91:9)
7
8
Conventional catalyst screening and experimental optimiza-
tion led to two experimental procedures delivering both dihy-
droquinolone isomers in good and synthetically useful regiose-
lectivities from the same substrate. However, a fundamental
mechanistic understanding of the origin of this behavior is of
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
3
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
cyclometalation to form I1, a styrene molecule first associates
with the catalyst (I1!I2) to give one of the four association
complexes I2-B1–I2-B4. Then, the olefin migratory inserts into
the five-membered rhodacycle (I2!I3), a process requiring
a significant amount of energy to overcome the relatively high
transition-state barrier TS1.[11] Following olefin insertion, I4 is
formed by dissociation of the Rh atom from the benzenoid
functional group. In some cases, an additional isomerization
step involving the seven-membered ring (I4!I5) is necessary
before the NꢀO bond linking the seven-membered ring with
the internal oxidant O-Boc group is broken (I5!I6). The cleav-
age of the NꢀO bond (TS2) is a step that also likely influences
the final product ratio. Following dissociation of the O-Boc
group, reductive elimination resulting in formation of a new
CꢀN bond (I6!I7) is also associated with a modest TS barrier
TS3. Formation of the six-membered ring (I7) is the irreversible
step leading to final product formation of either 4-aryl dihy-
droisoquinolone 4aa or 3-aryl dihydroisoquinolone 3aa. Closer
examination of the free energy profiles associated with each of
the four isomers of I2 distinguishes the most favorable path-
ways leading to the experimentally observed products.
Figure 1 (see the Supporting Information for quantitative
values) shows the reaction pathway energetics for the annula-
tion process with the CpCy ligand. Although the pathways cor-
responding to the different initial orientations of the styrene,
defined as B1–B4 (Scheme 2), do overlap at times, it is clear
from the free energy of the initial association step I1!I2 that
the I2-B4Cy is quite unfavorable owing to the large displace-
ment of the O-Boc group upon styrene association, meaning
Table 2. (Continued)
Entry
Conditions A
Conditions B
9
3da, 51% (94:6)
4da, 58% (90:10)
4ea, 68% (90:10)
10
11
3ea, 63% (92:8);
73% (>99:1)[b]
3 fa, 56% (92:8);
75% (>99:1)[b]
4 fa, 61% (88:12)
[a] Conditions A: 1 (0.10 mmol),
2
(0.20 mmol), Rh2 (5 mol%), (BzO)2
(5 mol%), 0.2m in CH2Cl2, 238C, 14 h; conditions B: 1 (0.10 mmol), 2
(0.20 mmol), [Cp*Rh(OBz)2] (5 mol%), 0.2m in CF3CH2OH, 238C, 14 h;
yields of the isolated major isomer; ratio of 4/3 was determined by
1H NMR spectroscopy of the crude product; [b] with 5 mol% Rh1 instead
of Rh2.
high interest and was therefore subsequently examined
through computational investigations.
Regioselectivity computational investigations
The starting point for all the different mechanistic
options giving products 3 or 4 is the cyclometalated
intermediate I1, which is commonly accepted to be
formed via the concerted metalation–deprotonation
(CMD)
CꢀH
activation
pathway
of
1 (Scheme 3).[11,22,23] The critical step is the coordina-
tion of the substituted olefin to I2 in four different
possible orientations, causing a competition between
different pathways leading to the two final products.
This situation is prototypical of a wide range of reac-
tions involving cyclometalated organometallic inter-
mediates reacting with acceptor moieties. Each of
the four associated isomeric styrene complexes (B1–
B4) formed by the different coordination modes then
proceeds through structurally similar, yet energetical-
ly unique, reaction pathways, finally leading to
a 4aa/3aa ratio based on the specific reaction ener-
getics and their associated kinetics (Scheme 3).
The ultimate goal is to understand and enhance
the factors responsible for the selectivity using chiral
Cp ligands for the development of future enantiose-
lective reactions. For this task, we first must be able
to describe the reaction free-energy profile and suc-
cessfully predict the trend for the major/minor prod-
uct ratio. The detailed reaction pathway involving the
RhIII catalyst with the four possible approaches of sty-
Scheme 3. Reaction pathway for CꢀH activation involving a RhIII catalyst with styrene
rene (B1–B4) is depicted in Scheme 3. After the initial and initial isomers of I2.
&
&
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
4
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
Figure 1. Reaction free-energy profile for the B1Cy–B4Cy isomers in the annulation reaction using the CpCy ligand. Computations at the PBE0[17]-dDsC[18]/TZ2P//
wB97X-D[19]/def2-SVP level including methanol implicit solvent using COSMO-RS.[20] See the Supporting Information for full computational details.[21]
that I2-BCy conformers should be populated by a mixture of I2-
B1Cy, I2-B2Cy, and I2-B3Cy, with I2-B2Cy having the highest con-
centration. However, the first TS barrier (TS1), associated with
migratory insertion of the olefin to form the seven-membered
rhodacycle, clearly identifies B3Cy as the kinetically preferred
pathway, since the TS1-B3Cy barrier is nearly 2.5 kcalmolꢀ1
(1 kcal=4.184 kJ) lower than the B1Cy conformer, its closest
competitor. Moreover, the resulting intermediate I3-B3Cy is also
thermodynamically favored relative to the other I3 isomers.
Moving forward from I3, the pathways of each conformer
(B1Cy–B4Cy) closely mirror one another. Once the open seven-
membered ring is formed (I4) the B1Cy and B3Cy pathways un-
dergo an additional isomerization step while B2Cy and B4Cy
move directly to the TS associated with NꢀO bond cleavage
(TS2). This TS barrier represents the final major point at which
product differentiation may occur. However, the barrier heights
for each of the four conformers are nearly equivalent, thus no
single pathway is predicted to be more kinetically favorable
than another from this particular step. Proceeding further
along the reaction pathway, I6 represents a seven-membered
ring species where NꢀO cleavage has occurred and the inter-
nal oxidant is now dissociated. The reductive ring closure form-
ing the dihydroquinolone by creating a new CꢀN bond (I6!
I7), is associated with only a small TS barrier height (TS3).
Nonetheless, TS3-B1Cy and TS3-B3Cy are slightly favored over
TS3-B2Cy and TS3-B4Cy. However, no significant differentiation
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
5
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
in product formation is expected from this barrier since the
B1Cy isomer leads to 4aa and B3Cy to 3aa. Finally, the TS3!I7
step is strongly exergonic and identifies the B3Cy isomer as the
thermodynamically preferred product of the complete reaction
pathway.
isomer. Since no other significant differences exist in the ener-
getics of the various pathways, our computations indicate that
the RhIII catalyst bearing the CpCy ligand results in production
predominately of 3aa. This finding is in full agreement with
experimental observations (Table 1, entry 3).
Considering the overall reaction free energy profile, the I2-
B3Cy isomer (Figure 1, blue lines) results in the largest contribu-
tion to the final products formed, as this conformer is the
most favorable from both a kinetic and thermodynamic per-
spective. The most influential step resulting in the preferred
formation of 3aa over 4aa likely is the depressed barrier
height associated with overcoming TS1-B3Cy for the I2-B3Cy
In contrast to the straightforward determination of a pre-
ferred product from the CpCy ligand, the energetics associated
with the Cp* ligand contain more ambiguity surrounding final
product selectivity (Figure 2). As with the CpCy ligand, the
B4Cp* pathway can be eliminated directly based on a relatively
unfavorable styrene association step (I1!I2) and a significantly
higher TS1 barrier height relative to all other isomers (for more
Figure 2. Reaction free-energy profile for the B1Cp*–B4Cp* isomers in the annulation reaction using the Cp* ligand. Computations at the PBE0-dDsC/TZ2P//
wB97X-D/def2-SVP level including methanol implicit solvent using COSMO-RS.
&
&
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
6
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
details, see the Supporting Information). Differentiating the re-
maining isomers is more challenging. Most likely, a mixture of
the B1Cp*, B2Cp*, and B3Cp* isomers exist after styrene associa-
tion, with B1Cp* having the highest concentration. However,
the free energy needed to overcome the TS1 barrier is smaller
for B3Cp* than for B1Cp*, which would ultimately result in for-
mation of a greater amount of 3aa, contradicting experimental
findings. Because the energetic picture surrounding the pre-
liminary steps does not clearly indicate a preferred isomer, the
later intermediates and TS barriers must be analyzed in more
detail to uncover the preference for production of 4aa.
Indeed, the lower barrier height of TS2-B1Cp* (corresponding
to separation of the final product from the O-Boc group) in
comparison to TS2-B3Cp* may be the origin of the preferred
formation of 4aa. The TS3 barrier height could also further dif-
ferentiate product formation in favor of the B1Cp* isomer and
4aa. However, this preference contrasts the final reaction step
(TS3!I7) that indicates B1Cp* as the least stable conformer.
Clearly, in the case of the Cp* ligand, even predicting the
major and minor products based on reaction pathway thermo-
dynamics alone is an extremely difficult task. As such, more re-
fined techniques are needed to uncover the desired informa-
tion about preferred product formation.
and styrene (0.4m), while concentrations of all intermediates
and products were set to zero. The movement between differ-
ent reaction intermediates is always associated with overcom-
ing a transition state barrier. However, only certain TS barriers
differ sufficiently between the various isomers and are large
enough to affect the ratio of final products. For the two ligand
systems studied here, TS1 (migratory olefin insertion), TS2 (Nꢀ
O cleavage), and TS3 (reductive CꢀN elimination) have been
identified as important TS barriers. All other transition states,
including the association of styrene to I1, and the steps involv-
ing isomerization of the seven-membered ring, were found to
minimally effect the final product ratios and, therefore, have
each been set to a constant value (4.0 kcalmolꢀ1 for barrier
heights not reported earlier) for each of the four isomers. The
results of the kinetic simulations displaying final product infor-
mation are shown in Figure 3. For the Cp* ligand (Figure 3a),
4aa is favorably produced at the reaction’s onset, with 3aa be-
coming dominant at intermediary times, and finally reaching
a 4aa/3aa ratio of 49:51 (Figure 3b). The origin of the initial
favorability of 4aa can be traced to the thermodynamically
preferred binding of styrene in a manner that produces more
4aa and less 3aa (I1-B1Cp*!I2-B1Cp*). However, the lower TS1
barrier height that leads to formation of 3aa (I2-B3Cp*!TS1-
B3Cp*) becomes more important as the reaction proceeds. Fi-
nally, the near equivalence in the 3aa and 4aa products likely
results from the relatively similar TS barrier heights (particularly
TS1) for the B1Cp* and B3Cp* isomers, each of which are ex-
pected to be the major contributors to their respective final
products (Figure 2).
Owing to the murky picture surrounding the thermodynam-
ics and kinetics of the Cp* ligand, it is desirable to proceed
beyond the free-energy reaction pathway modeling exempli-
fied in Figure 1 and Figure 2, and directly access kinetic infor-
mation that reveals the final product ratios. The coupling of
thermodynamic data derived from either quantum chemical
methods or experiment and transition state theory (TST) is
a tool that has been employed to access kinetic information in
various domains traditionally associated with chemical engi-
neering. When used together, these methods provide a means
to estimate the concentrations of different product and inter-
mediate species over time. For example, approaches based on
TST have assisted in the modeling of combustion[24] and poly-
merization[25] processes. Their use in catalysis and organometal-
lic chemistry, however, is scarce. Reiher and co-workers recent-
ly used TST to determine the kinetics of various [Fe]-hydroge-
nase model compounds,[26] whereas ligand addition reactions
to an iron carbonyl complex have been examined by George,
Harvey and co-workers.[27] Moreover, Kozuch and Shaik used
TST to develop a new model to calculate turnover frequency
of catalytic cycles.[28] Applying a TST model to the cases involv-
ing the CpCy and Cp* ligands should provide direct access to
crucial information concerning the final product ratios.
For the CpCy ligand (Figure 3c), the kinetic picture confirms
predictions made through examination of the reaction free
energy profile. The B3Cy pathway is consistently seen to have
intermediates amongst the most thermodynamically stable
and the lowest TS barrier heights (Figure 1). As expected, these
favorable thermodynamics translate directly into a clear prefer-
ence for the production of 3aa when the CpCy ligand is used.
Our simulations estimate the final 4aa/3aa ratio to be 24:76
(Figure 3d).
Immediately evident from the use of these kinetic models is
the distinct advantage of directly accessing major and minor
product ratios that arise during the catalytic process. This mir-
rored the experimentally observed trend in selectivity. Howev-
er, an exact quantitative reproduction of the observed product
ratios was not anticipated as even minimal changes in the re-
action free-energy profile (in the sub-kcalmolꢀ1 range) enor-
mously impact the final product ratios. For instance, decreas-
ing the height of the TS1 barriers along 4aa-forming pathways
(Figure 2, green and black lines) while increasing the same bar-
rier along the 3aa-forming pathways (Figure 2, purple and
blue lines) by only 0.1 kcalmolꢀ1 shifts the major product from
3aa to 4aa, which qualitatively matches experimental observa-
tion (4aa/3aa ratio of 57:43). Moving the same barrier heights
by 0.5 kcalmolꢀ1 reproduces, nearly exactly (83:17), the experi-
mentally observed 84:16 ratio in methanol. This dramatic de-
pendence and sensitivity of the final product ratios on the TS1
barrier height is further illustrated in Table S7 (see the Support-
ing Information). Thus computational treatments, capable of
The first step in evaluating the kinetic data involves defining
the rate constants and establishing the systems of ordinary dif-
ferential equations that govern the concentrations of all spe-
cies present in the reaction scheme. Forty-four rate constants
are defined for the transitions between starting, intermediate,
and final products (see the Supporting Information). Based on
these rate constants, a series of 24 ordinary differential equa-
tions can be constructed (see the Supporting Information) that
measure the change in concentration of the 24 chemical spe-
cies potentially present in a reaction mixture over time. Our ki-
netic simulations initially set concentrations for I1 (0.001m)
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
7
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
Figure 3. Evolution of final product species over time. Panels (a) and (c) represent the product evolution for the Cp* and CpCy ligands, respectively, while
panels (b) and (d) provide only the late time steps permitting a visualization of product ratios.
providing reliable sub-kcalmolꢀ1 accuracy of electronic ener-
gies along with prohibitively expensive treatments of the dy-
namic fluctuation of explicit solvent molecules would be nec-
essary to achieve reaction free energies that reproduce the
exact product ratios observed in experiment. Unfortunately,
such protocols are currently out of reach. Despite the quantita-
tive shortcoming, our computational analysis does clearly iden-
tify the RhIII catalyst bearing the Cp* ligand as a better produc-
er of 4aa than the CpCy ligand, whereas the latter is a superior
choice if 3aa is desired, matching the experimentally observed
trend (Table 1). Arriving at this conclusion would not be possi-
ble simply by examining the reaction free energy profile
(Figure 2).
Conclusion
We have shown that RhIII catalysts equipped with tailored cy-
clopentadienyl ligands allow high catalyst control for regiodi-
vergent pathways for CꢀH functionalizations of aryl hydroxa-
mates with styrenes. While a 1,2-disubstituted cyclopentadienyl
ligand—CpCy—selectively gives 3-aryl dihydroisoquinolones,
the Cp*-bearing complex selectively provides previously totally
inaccessible and synthetically valuable 4-aryl dihydroisoquino-
lones. Both the steric bulk of the cyclopentadienyl group of
the ligand and the Boc substituent of the built-in hydroxamate
oxidant play critical and synergistic roles in the regioselectivity-
determining alignment of the styrene acceptor. This demon-
strates the power of ligand-controlled regiochemistry enabling
access to divergent products from the same substrate. Subse-
quent computational investigations revealed that the exact co-
&
&
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
8
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
ordination mode of the olefin to the cyclometalated RhIII inter-
mediate is a key determinant for the different product-forming
pathways. Moreover, we demonstrated that combining state-
of-the-art density functional computations with a kinetic
model based on transition state theory provide access to re-
gioselectivity trends observed in the catalytic process. Rather
than relying upon “best guess” estimates determined by ex-
amination of crisscrossing reaction free energy profiles, the
monitoring of reaction species concentrations provided by ki-
netic simulations permit direct examination of the effects
chemical modifications have on catalytic efficiency. Modeling
of this type could become an essential tool in the computa-
tional design and in silico analysis of new catalysts by permit-
ting direct access to relative regioselectivity.
1554; f) N. Kuhl, N. Schrçder, F. Glorius, Adv. Synth. Catal. 2014, 356,
1443.
[5] For representative examples with alkynes, see: a) K. Ueura, T. Satoh, M.
Miura, Org. Lett. 2007, 9, 1407; b) J. Chen, G. Song, C.-L. Pan, X. Li, Org.
Lett. 2010, 12, 5426; c) T. K. Hyster, T. Rovis, J. Am. Chem. Soc. 2010, 132,
10565; d) S. Rakshit, F. W. Patureau, F. Glorius, J. Am. Chem. Soc. 2010,
132, 9585; e) J. Chen, G. Song, C.-L. Pan, X. Li, Org. Lett. 2010, 12, 5426;
f) P. C. Too, Y.-F. Wang, S. Chiba, Org. Lett. 2010, 12, 5688; g) N. Guimond,
C. Gouliaras, K. Fagnou, J. Am. Chem. Soc. 2010, 132, 6908; h) G. Song,
D. Chen, C.-L. Pan, R. H. Crabtree, X. Li, J. Org. Chem. 2010, 75, 7487; i) S.
Mochida, N. Umeda, K. Hirano, T. Satoh, M. Miura, Chem. Lett. 2010, 39,
744; j) X. Li, M. Zhao, J. Org. Chem. 2011, 76, 8530; k) H. Wang, C. Groh-
mann, C. Nimphius, F. Glorius, J. Am. Chem. Soc. 2012, 134, 19592;
l) M. V. Pham, B. Ye, N. Cramer, Angew. Chem. Int. Ed. 2012, 51, 10610;
Angew. Chem. 2012, 124, 10762; m) J. Jayakumar, J. Parthasarathy, C.-H.
Cheng, Angew. Chem. Int. Ed. 2012, 51, 197; Angew. Chem. 2012, 124,
201; n) J. Karthikeyan, R. Haridharan, C.-H. Cheng, Angew. Chem. Int. Ed.
2012, 51, 12343; Angew. Chem. 2012, 124, 12509; o) N. Quinones, A.
Seoane, R. Garcia-Fandino, J. J. Mascarenas, M. Gulias, Chem. Sci. 2013,
4, 2874; p) J. M. Neely, T. Rovis, J. Am. Chem. Soc. 2013, 135, 66; q) K.
Muralirajan, C.-H. Cheng, Chem. Eur. J. 2013, 19, 6198; r) J. Seo, Y. Park, I.
Jeon, T. Ryu, S. Park, P. H. Lee, Org. Lett. 2013, 15, 3358; s) Y. Unoh, Y. Ha-
shimoto, D. Takeda, K. Hirano, T. Satoh, M. Miura, Org. Lett. 2013, 15,
3258; t) S.-C. Chuang, P. Gandeepan, C.-H. Chen, Org. Lett. 2013, 15,
5750; u) A. Seoane, N. Casanova, N. Quinones, J. L. Mascarenas, M.
Gulias, J. Am. Chem. Soc. 2014, 136, 834.
Acknowledgements
This work is supported by the European Research Council
under the European Community’s Seventh Framework Pro-
gram (FP7 2007-2013)/ERC Grant agreements no 257891 and
the Swiss National Science Foundation (no. 137666 and
137529).
[6] For representative examples with allenes, see: a) H. Wang, F. Glorius,
Angew. Chem. Int. Ed. 2012, 51, 7318; Angew. Chem. 2012, 124, 7430;
b) R. Zeng, C. Fu, S. Ma, J. Am. Chem. Soc. 2012, 134, 9597; c) B. Ye, N.
Cramer, J. Am. Chem. Soc. 2013, 135, 636; d) R. Zeng, S. Wu, C. Fu, S.
Ma, J. Am. Chem. Soc. 2013, 135, 18284; e) T.-J. Gong, W. Su, Z.-J. Liu,
W.-M. Cheng, B. Xiao, Y. Fu, Org. Lett. 2014, 16, 330.
Keywords: density functional calculations
· homogeneous
catalysis · ligand control · rhodium · transition state theory
[7] For representative examples with olefins, see: a) F. Wang, G. Song, X. Li,
Org. Lett. 2010, 12, 5430; b) S. Rakshit, C. Grohmann, T. Besset, F. Glorius,
J. Am. Chem. Soc. 2011, 133, 2350; c) N. Guimond, S. I. Gorelsky, K.
Fagnou, J. Am. Chem. Soc. 2011, 133, 6449; d) Z. Shi, N. Schroer, F. Glo-
rius, Angew. Chem. Int. Ed. 2012, 51, 8092; Angew. Chem. 2012, 124,
8216; e) C. Zhu, J. R. Falck, Chem. Commun. 2012, 48, 1674; f) J. M.
Neely, T. Rovis, J. Am. Chem. Soc. 2013, 135, 66; g) S. Cui, Y. Zhang, Q.
Wu, Chem. Sci. 2013, 4, 3421; h) Z. Shi, C. Grohmann, F. Glorius, Angew.
Chem. Int. Ed. 2013, 52, 5393; Angew. Chem. 2013, 125, 5503; i) T. A.
Davis, T. K. Hyster, T. Rovis, Angew. Chem. Int. Ed. 2013, 52, 14181;
Angew. Chem. 2013, 125, 14431; j) B. Ye, P. A. Donets, N. Cramer, Angew.
Chem. Int. Ed. 2014, 53, 507; k) Z. Shi, M. Boultadakis-Arapinis, D. C.
Koester, F. Glorius, Chem. Commun. 2014, 50, 2650.
[1] a) Selectivity in Catalysis (Eds.: M. E. Davis, S. L. Suib), ACS Symposium
Series 517, American Chemical Society, Washington, 1993; b) R. S. Ward,
Selectivity in Organic Synthesis, Wiley, New York, 1999.
[2] For recent selected examples involving metal catalysis, see: a) Z. Ke, S.
Abe, T. Ueno, K. Morokuma, J. Am. Chem. Soc. 2011, 133, 7926; b) M.
Garcꢃa-Melchor, S. I. Gorelsky, T. K. Woo, Chem. Eur. J. 2011, 17, 13847;
c) S. Mazumder, D. Shang, D. E. Negru, M.-H. Baik, P. A. Evans, J. Am.
Chem. Soc. 2012, 134, 20569; d) S. I. Gorelsky, D. Lapointe, K. Fagnou, J.
Org. Chem. 2012, 77, 658; e) X. Zhang, J. Wang, H. Zhao, H. Zhao, J.
Wang, Organometallics 2013, 32, 3529; f) X. Hong, B. M. Trost, K. N.
Houk, J. Am. Chem. Soc. 2013, 135, 6588; g) X. Xu, P. Liu, X.-Z. Shu, W.
Tang, K. N. Houk, J. Am. Chem. Soc. 2013, 135, 9271; h) I. A. Sanhueza,
A. M. Wagner, M. S. Sanford, F. Schoenebeck, Chem. Sci. 2013, 4, 2767;
i) G.-J. Cheng, Y.-F. Yang, P. Liu, P. Chen, T.-Y. Sun, G. Li, X. Zhang, K. N.
Houk, J.-Q. Y, Y.-D. Wu, J. Am. Chem. Soc. 2014, 136, 894; j) S. H. Park, J.
Kwak, K. Shin, J. Ryu, Y. Park, S. Chang, J. Am. Chem. Soc. 2014, 136,
2492.
[8] For seminal reports of hydroxamate ester directing groups for CꢀH
functionalizations, see: a) L. E. Fisher, J. M. Caroon, S. Jahangir, S. R. Sta-
bler, S. Lundberg, J. M. Muchowski, J. Org. Chem. 1993, 58, 3643; b) D.-
H. Wang, M. Wasa, R. Giri, J.-Q. Yu, J. Am. Chem. Soc. 2008, 130, 7190.
[9] B. Ye, N. Cramer, Science 2012, 338, 504.
[10] M. Presset, D. Oehlrich, F. Rombouts, G. A. Molander, Org. Lett. 2013, 15,
1528.
[3] For selected recent reviews on CꢀH functionalization, see: a) J. Wencel-
Delord, F. Glorius, Nat. Chem. 2013, 5, 369; b) J. Yamaguchi, A. D. Yama-
guchi, K. Itami, Angew. Chem. Int. Ed. 2012, 51, 8960; Angew. Chem.
2012, 124, 9092; c) S. H. Cho, J. Y. Kim, J. Kwak, S. Chang, Chem. Soc.
Rev. 2011, 40, 5068; d) J. Wencel-Delord, T. Droge, F. Liu, F. Glorius,
Chem. Soc. Rev. 2011, 40, 4740; e) C. Liu, H. Zhang, W. Shi, A. Lei, Chem.
Rev. 2011, 111, 1780; f) J. Le Bras, J. Muzart, Chem. Rev. 2011, 111, 1170;
g) T. W. Lyons, M. S. Sanford, Chem. Rev. 2010, 110, 1147; h) D. A. Colby,
R. G. Bergman, J. A. Ellman, Chem. Rev. 2010, 110, 624; i) F. Kakiuchi, T.
Kochi, Synthesis 2008, 3013; j) D. Alberico, M. E. Scott, M. Lautens,
Chem. Rev. 2007, 107, 174; k) X. Chen, K. M. Engle, D.-H. Wang, J.-Q. Yu,
Angew. Chem. Int. Ed. 2009, 48, 5094; Angew. Chem. 2009, 121, 5196;
l) L. Ackermann, R. Vicente, A. R. Kapdi, Angew. Chem. Int. Ed. 2009, 48,
9792; Angew. Chem. 2009, 121, 9976.
[11] L. Xu, Q. Zhu, G. Huang, B. Cheng, Y. Xia, J. Org. Chem. 2012, 77, 3017.
[12] For examples on the differences between various substituted Cp li-
gands in RhIII catalysis, see: a) Y. Shibata, K. Tanaka, Angew. Chem. Int.
Ed. 2011, 50, 10917; Angew. Chem. 2011, 123, 11109; b) T. K. Hyster, T.
Rovis, Chem. Commun. 2011, 47, 11846; c) T. K. Hyster, T. Rovis, Chem.
Sci. 2011, 2, 1606; d) N. Umeda, H. Tsurugi, T. Satoh, M. Miura, Angew.
Chem. Int. Ed. 2008, 47, 4019; Angew. Chem. 2008, 120, 4083; e) T. Uto,
M. Shimizu, K. Ueura, H. Tsurugi, T. Satoh, M. Miura, J. Org. Chem. 2008,
73, 298; f) M. Shimizu, K. Hirano, T. Satoh, M. Miura, J. Org. Chem. 2009,
74, 3478.
[13] a) M. P. Huestis, L. Chan, D. R. Stuart, K. Fagnou, Angew. Chem. Int. Ed.
2011, 50, 1338; Angew. Chem. 2011, 123, 1374; b) T. K. Hyster, T. Rovis,
Synlett 2013, 24, 1842.
[4] For overviews on Cp*RhIII-catalyzed CꢀH functionalizations, see: a) T.
Satoh, M. Miura, Chem. Eur. J. 2010, 16, 11212; b) F. W. Patureau, J.
Wencel-Delord, F. Glorius, Aldrichimica Acta 2012, 45, 31; c) G. Song, F.
Wang, X. Li, Chem. Soc. Rev. 2012, 41, 3651; d) C. Zhu, R. Wang, J. R.
Falck, Chem. Asian J. 2012, 7, 1502; e) S. Chiba, Chem. Lett. 2012, 41,
[14] T. K. Hyster, L. Knçrr, T. R. Ward, T. Rovis, Science 2012, 338, 500.
[15] J. A. Gladysz, B. J. Boone, Angew. Chem. 1997, 109, 566; Angew. Chem.
Int. Ed. Engl. 1997, 36, 550.
[16] P. G. Gassman, J. W. Mickelson, J. R. Sowa Jr., J. Am. Chem. Soc. 1992,
114, 6942.
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
9
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
[17] a) J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 1996, 77, 3865;
b) C. Adamo, V. Barone, J. Chem. Phys. 1999, 110, 6158.
[23] The evaluated transition states and intermediate are constraints on the
CMD mechanism as this is the generally accepted pathway for this type
of transformation.
[24] a) J. A. Miller, M. J. Pilling, J. Troe, Proc. Int. Conf. ESNA Working Group on
“Waste Irradiation” Proc. Combust. Inst. 2005, 30, 43; b) A. G. Vandeputte,
M. K. Sabbe, M.-F. Reyniers, V. van Speybroeck, M. Waroquier, G. B.
Marin, J. Phys. Chem. A 2007, 111, 6772.
[18] a) S. N. Steinmann, C. Corminboeuf, J. Chem. Theory Comput. 2010, 6,
1990; b) S. N. Steinmann, C. Corminboeuf, J. Chem. Phys. 2011, 134,
044117; c) S. N. Steinmann, C. Corminboeuf, Chimia 2011, 65, 240;
d) S. N. Steinmann, C. Corminboeuf, J. Chem. Theory Comput. 2011, 7,
3567.
[25] a) E. I. Izgorodina, M. L. Coote, Chem. Phys. 2006, 324, 96; b) X. Yu, J.
Pfaendtner, L. Broadbelt, J. Phys. Chem. A 2008, 112, 6772.
[26] A. R. Finkelmann, M. T. Stiebritz, M. Reiher, J. Phys. Chem. B 2013, 117,
4806.
[27] M. Besora, J.-L. Carreꢄn-Macedo, A. J. Cowan, M. W. George, J. N. Harvey,
P. Portius, K. L. Ronayne, X.-Z. Sun, M. Towrie, J. Am. Chem. Soc. 2009,
131, 3583.
[19] a) J.-D. Chai, M. Head-Gordon, J. Chem. Phys. 2008, 128, 084106; b) J.-D.
Chai, M. Head-Gordon, Phys. Chem. Chem. Phys. 2008, 10, 6615.
[20] A. Klamt, WIREs Comput. Mol. Sci. 2011, 1, 699.
[21] Optimizations using the M06 density functional produced the same
trends as the wB97X-D density functional. See the Supporting Informa-
tion for details.
[22] a) D. Garcꢃa-Cuadrado, P. de Mendoza, A. A. C. Braga, F. Maseras, A. M.
Echavarren, J. Am. Chem. Soc. 2007, 129, 6880; b) D. Lapointe, K.
Fagnou, Chem. Lett. 2010, 39, 1118; c) L. Ackermann, Chem. Rev. 2011,
111, 1315; d) S. I. Gorelsky, Coord. Chem. Rev. 2013, 257, 153.
[28] S. Kozuch, S. Shaik, Acc. Chem. Res. 2011, 44, 101.
Received: July 22, 2014
Published online on && &&, 0000
&
&
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
10
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
FULL PAPER
&
CꢀH Functionalization
M. D. Wodrich, B. Ye, J. F. Gonthier,
C. Corminboeuf,* N. Cramer*
&& – &&
RhIII-catalyzed directed CꢀH functional-
ization of aryl hydroxamates enables
the selective formation of regioisomeric
3-aryl or 4-aryl dihydroisoquinolones
with two different catalytic systems. The
different selectivities are examined
using DFT and transition state theory.
The stabilities of the rhodium complex
adducts and migratory insertion activa-
tion barriers are key factors for the re-
giodivergent pathways (see scheme;
rs=regioselectivity).
Ligand-Controlled Regiodivergent
Pathways of Rhodium(III)-Catalyzed
Dihydroisoquinolone Synthesis:
Experimental and Computational
Studies of Different Cyclopentadienyl
Ligands
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
11
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ