The scope and mechanism of palladium-catalysed Markovnikov alkoxycarbonylation of alkenes

Article abstract of DOI:10.1038/nchem.2586

Hydroesterification reactions represent a fundamental type of carbonylation reaction and constitute one of the most important industrial applications of homogeneous catalysis. Over the past 70 years, numerous catalyst systems have been developed that allow for highly linear-selective (anti-Markovnikov) reactions and are used in industry to produce linear carboxylates starting from olefins. In contrast, a general catalyst system for Markovnikov-selective alkoxycarbonylation of aliphatic olefins remains unknown. In this paper, we show that a specific palladium catalyst system consisting of PdX₂/N-phenylpyrrole phosphine (X, halide) catalyses the alkoxycarbonylation of various alkenes to give the branched esters in high selectivity (branched selectivity up to 91%). The observed (and unexpected) selectivity has been rationalized by density functional theory computation that includes a dispersion correction.

Full text of DOI:10.1038/nchem.2586

ARTICLES
PUBLISHED ONLINE: 5 SEPTEMBER 2016 | DOI: 10.1038/NCHEM.2586
The scope and mechanism of palladium-catalysed
Markovnikov alkoxycarbonylation of alkenes
Haoquan Li, Kaiwu Dong, Haijun Jiao, Helfried Neumann, Ralf Jackstell and Matthias Beller
*
Hydroesterification reactions represent a fundamental type of carbonylation reaction and constitute one of the most
important industrial applications of homogeneous catalysis. Over the past 70 years, numerous catalyst systems have been
developed that allow for highly linear-selective (anti-Markovnikov) reactions and are used in industry to produce linear
carboxylates starting from olefins. In contrast, a general catalyst system for Markovnikov-selective alkoxycarbonylation
of aliphatic olefins remains unknown. In this paper, we show that a specific palladium catalyst system consisting of
PdX2/N-phenylpyrrole phosphine (X, halide) catalyses the alkoxycarbonylation of various alkenes to give the branched
esters in high selectivity (branched selectivity up to 91%). The observed (and unexpected) selectivity has been
rationalized by density functional theory computation that includes a dispersion correction.
wing to the perfect atom economy, the addition of vinylarenes and vinylacetate23–32. In this context, Markovinkov-
HNu (Nu = halogen, CR3, CN, OH, NR2 and so on) to selective alkoxycarbonylations of easily available bulk olefins, such
O
carbon–carbon multiple bonds is one of the most-desired as butenes, hexenes, octenes and so on, are basically unknown,
transformations in organic synthesis1. Using abundant and inexpen- even though the potential products that would arise from such reactions
sive starting materials a number of industrial processes take have broad utility in the fragrance and flavour industry33,34 as
advantage of these methods. Well-recognized examples include well as in the life-science industries (Fig. 1b). Moreover, branched
hydroamination and hydration, as well as others. In general, most aliphatic esters are used as important intermediates in organic
of the electrophilic addition reactions of alkenes follow the so- synthesis35–38. In view of a more-sustainable production for such
called Markovnikov rule (Fig. 1a). In the polar addition reaction of compounds, the development of a general and selective catalytic
Nu–E to alkenes, the superior stability of the resulting secondary car- methodology is highly desirable.
bocation RCH+CH2E compared with the primary carbocation
Based on our long-standing interest in carbonylation reactions,
RCEHCH2+ controls the regioselectivity of the process and generates we report herein our recent investigations on a general and universal
the corresponding branched ‘Markovnikov product’ after nucleophi- method for the Markovinkov alkoxycarbonylation of aliphatic
lic attack of the nucleophile2. Interestingly, reactions of alkenes that 1-alkenes. One of the prerequisites for the Markovnikov selective
are catalysed by transition metals, such as hydrocyanation, hydro- carbonylation of alkenes (≥C5) is to control the isomerization
silylation and others3–6, and carbonylation reactions (hydroformyl- reaction of the alkene (Fig. 1c)39–42. One obvious reason is the
ation, alkoxycarbonylation and aminocarbonylation) offer the variety of esters that arise from the carbonylation of different
possibility to generate the corresponding linear products. In these internal alkenes. For example, the simple 1-octene itself might
latter reactions the selective formation of the Markovnikov products form four different regioisomeric C9 esters (3aa-1 to 3aa-4). In
is difficult7. In fact, regioselective alkene functionalization continues the case of di- and multisubstituted olefins even more esters can
to be an exceedingly challenging goal for (homogeneous) catalysis8. be formed. To avoid the formation of such mixtures, we anticipated
In general, carbonylation reactions constitute one of the most that an ‘acid-free’ catalyst system could prevent olefin-isomerization
important alkene-transformation reactions with regard to the reactions and keep the terminal alkene as a major component in the
production scale9–12. In this context, the direct carbonylation of abun- reaction media.
dantly available alkenes to give aldehydes and carboxylic esters,
so-called hydroformylation (∼10 million tons per year)13, and
Results and discussion
alkoxycarbonylation (multimillion tons per year) are two key trans-
formations for the chemical industry14,15. As pointed out above, the
linear product (anti-Markovnikov product) is usually favoured for
the unactivated alkene carbonylations developed thus far. In fact,
continuing activities over the past four decades, mainly based on
the development of phosphorus ligands, established efficient and
highly linear-selective catalytic systems for hydroformylations and
alkoxycarbonylations of aliphatic alkenes, which are applied in
Catalytic experiments. At the beginning of our study, 1-octene
(1 mmol) and methanol (1 mmol) were chosen as the benchmark
substrates. To control the regioselectivity of this transformation,
carbonylation experiments in the presence of PdCl2 (1 mol%) and
various standard and specifically made phosphine ligands were
performed (Fig. 2). It is well known that bidentate phosphine
ligands preferentially form linear esters from both internal and
terminal unactivated olefins16,43,44
. Hence, initially different
industry as well16–18
.
monodentate phosphine ligands were tested with our model
substrates. When triarylphosphine ligands L1, L2 and L4 were
applied, almost equal amounts of branched and linear products
were obtained. To our surprise, in the presence of L3 (tris-
(2-methoxyphenyl)phosphine) a yield of 74% of C9 esters with
73% selectivity for the desired branched product was obtained.
The formation of the branched product for both hydroformylation
and alkoxycarbonylation reactions is more challenging because of
the increase in steric congestion for the secondary carbon–metal
intermediates19–22. Branch-selective alkoxycarbonylations can be
predominant only with special substrates, such as allenes,
Leibniz-Institut für Katalyse e. V. an der Universität Rostock, Albert-Einstein Straße 29a, Rostock 18059, Germany. e-mail: matthias.beller@catalysis.de
*
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
1
© 2016 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.2586
ARTICLES
O
O
a
b
O
Ph
O
Markovnikov
Nu
Anti-Markovnikov
E
Fruity, herbaceous, floral
O
Citrus, strawberry
Polar addition
NuE
E
Nu
+
+
R
R
O
Favoured
CONu
Not favoured
H
O
Ph
R
O
R = alkyl
Metal catalyst
CO, NuH
Apple, plum
O
Fruity, floral
H
CONu
R
R
Carbonylation
OH
Not favoured
Favoured
O
O
Fruity manzanate
Butter, cheese, creamy
c
Pd-catalysed
isomerization
C3H7
C3H7
C4H9
C4H9
C4H9
Pd-catalysed
carbonylation
CO, MeOH
COOMe
CO, MeOH
CO, MeOH
COOMe
CO, MeOH
C4H9
COOMe
C4H9
C4H9
C4H9
COOMe
3aa-3
3aa-1
3aa-2
3aa-4
Figure 1 | Challenges and application potentials for the palladium-catalysed Markovnikov alkoxycarbonylation of alkenes. a, Comparison of the general
product distribution for the polar addition and carbonylation reactions of alkenes. b, Selected examples of 2-methylcarboxylates and their organoleptic
properties. c, Possible products from the palladium-catalysed methoxycarbonylation of 1-octene. The purple circles highlight the Markovnikov reaction centre.
1 mol% PdCl2,
3aa-1: 1-COOMe
4
2
COOMe
2 or 4 mol% ligand
3aa-2: 2-COOMe
3aa-3: 3-COOMe
3aa-4: 4-COOMe
+
MeOH
C4H9
C4H9
3
1
Toluene, 100 °C, 20 h,
CO (40 bar)
1a
2a
Me
O
PCy2
PCy2
i-Pr
PR3
Cy
P
PCy2
PCy2
i-Pr
P
L1: R = Ph, 15 (48)
N
L2: R = 2-furyl, 76 (54)
Cy
L3: R = 2-OMe-Ph, 74 (73)
i-Pr
L7, 9 (48)
L4, 87 (55)
L5, 71 (55)
L6, 42 (87)
L8, 29 (76)
L9, 29 (85)
N
PR2
OMe
PR2
N
PPh2
OMe
N
PCy2
N
N
O
PPh2
O
PPh2
L10: R = t-Bu, 10 (45)
L11: R = Ph, 66 (41)
L12: R = Cy, 49 (79)
L13: R = o-tol, 30 (75)
PPh2
PPh2
L14, 49 (82)
L15: R = Ph, 20 (80)
L16: R = Cy, 41 (89)
L17, 3 (61)
L18, 86 (33)
L19, 31 (13)
Figure 2 | Markovnikov alkoxycarbonylation of 1-octene using a range of different ligands. Reaction conditions: 1a (1.0 mmol), 2a (1.0 mmol), PdCl2 (1 mol
%), L1–L17 (4 mol%) or L18 or L19 (2 mol%), CO (40 bar), toluene (2 ml), 100 °C, 20 h. The yields (%) of the mixtures of all the C9 esters are based on
methanol as determined by GC analysis using isooctane as the internal standard. The branched selectivity = (3aa-2/(3aa-1 + 3aa-2 + 3aa-3 + 3aa-4)) and is
given in brackets (%). The ratios of isomers were determined by GC analysis. In all the cases shown here, insignificant amounts of 3aa-3 and 3aa-4
were observed. tol, toluene.
Apparently, the increased steric hindrance of the ligand caused by was observed with L5 PCy2Ph, which clearly shows that the biaryl
the ortho substitution prefers the formation of the desired function is important to obtain the desired branched selectivity.
branched product. Consequently, more-sterically hindered ligands, Subsequently, different commercially available cataCXium series
such as the biaryl ligands L14 and L17, were tested. Indeed, with ligands were tested46. To our delight, in the presence of the
the commercially available biaryl ligands that were reported to be methylfluorenylphosphine ligand L9, 85% selectivity was observed.
excellent ligands for several coupling reactions, modest-to-good Among all the N-phenyl–pyrrole-based ligands tested (L10–L13),
branched selectivity was observed45. For example, Mephos L6 and the cyclohexyl-substituted derivative L12 gave the best branched
Davephos L8 gave 42 and 29% yields and 87 and 76% selectivities, selectivity (79%). When the related ligand L14 was applied with the
respectively. For comparison, a branched selectivity of only 55% N-phenylindole backbone, a similar result was observed (49% yield
2
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
© 2016 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.2586
ARTICLES
Table 1 | Substrate scope for the Markovnikov alkoxycarbonylation of alkenes.
1 mol% PdCl2,
4 mol% L16
2
COOR
+
ROH
R
R
Toluene, 100 °C, 20 h,
CO (40 bar), 5 µL H2O
1
1x
2a R = Me
2d R = PhCH2
3xy-1: 1-COOR
3xy-2: 2-COOR
Entry*
Alkene†
Alcohol
Yield (%)‡
Branched selectivity (%)§
3xa-2/iso (%)§
1
1-octene
2-octene
1-pentene
1-decene
1a
2a
2a
2a
2a
88
50
81
84
97
81
96
77
98
98
2
3
4
1b
1c
1d
88
83
CN
5
1e
2d
86
85
>99
Cl
6
7
1f
2a
2a
93
71
83
87
>99
>99
Br
1g
OAc
8
9
1h
1i
2a
2a
97
90
84
83
96
96
COOMe
8
Ph
10
1j
2a
90
84
>99
Ph
11
1k
1l
2a
2d
82
78
>99
78
>99
>99
12
13
14
1m
1n
2d
2a
91
49
52
>99
99
88
*Reaction conditions: 1x (2 mmol), 2a (1 mmol), PdCl2 (0.01 mmol), cataCXium POMeCy (0.04 mmol), CO (40 bar), toluene, 100 °C, 5 µl H2O, 20 h. †Green shading, functional group leads to an even better
branched selectivity; yellow shading, functionality is well tolerated and leads to a good branched selectivity; red shading, functionality leads to a decreased branched selectivity. ‡Yield of isolated mixture of
iso- and n-esters based on the alcohol. §Branched selectivity and 3xy-2/iso ratios were determined by GC.
and 82% branched selectivity). Gratifyingly, with N-(o-MeO-phenyl)-
Based on these preliminary studies, the following conditions
pyrrole as the ligand backbone, L15 and the commercially available were used for further investigations: 1 mol% PdCl2, 4 mol% L16,
ligand L16 (cataCXium POMeCy) gave high regioselectivities 5 µl H2O in 2 ml toluene, CO (40 bar), 100 °C for 20 hours. Then,
(respectively, 20 and 41% yield, 80 and 89% branched selectivity). different alkenes were tested using either methanol or benzyl
Unfortunately, the analogue N-phenylimidazole ligand L17 gave alcohol (Table 1). Simple short- and long-chain aliphatic olefins,
only a low yield (3%). For comparison, both Xantphos L18 and for example 1-pentene or 1-decene, gave good yields and good
DPEphos L19 gave the linear product as the major product branched selectivity (81 and 88% yield, with 81 and 83% selectivity,
(branched selectivity 33 and 13%, respectively).
respectively (Table 1, entries 3 and 4)). With 2-octene as the starting
With L16 as the ligand of choice, we tried to improve the reactiv- substrate, 97% branched selectivity was obtained. Among the
ity further (Supplementary Table 1, entry 2). By adding 10 µl H2O, branched products, 76.2% 3aa-2, 23.0% 3aa-3 and 0.8% 3aa-4
which probably facilitates the formation of the active palladium were observed by gas chromatography (GC) analysis (Table 1,
hydride, the yield was improved to 63% with a slightly decreased entry 2). To the best of our knowledge, this constitutes the highest
branched selectivity (84%) and a 92% 3aa-2/iso ratio (branched branched regioselectivity reported for any non-functionalized aliphatic
selectivity indicates the ratio of iso-esters (3aa-2 + 3aa-3 + 3aa-4) olefin. Furthermore, an excellent functional group tolerance was
to all the generated C9 esters (3aa-1 + 3aa-2 + 3aa-3 + 3aa-4; observed. Alkenes that bear –CN, –Cl, –Br, –OAc and –COOMe
the 3aa-2/iso ratio is the selectivity of the Markovnikov product were shown to be compatible, and gave the corresponding ester in
2-carboxylate to all the iso-esters in the final product). Comparably, 71–97% yield with 83–87% branched selectivity (Table 1, entries 5–10).
with the addition of 5 mol% PTSA (p-toluenesulfonic acid monohy- Excellent branched selectivity (>99%) was observed for styrene
drate), the activity was promoted but a larger amount of the internal and the corresponding iso product was isolated with 82% yield
carbonylated products 3aa-3 and 3aa-4 was obtained (Supplementary (Table 1, entry 11). For steric reasons, the introduction of substitu-
Table 1, entry 3). Finally, by increasing the amount of olefin ents into the 3- or 4-position of a terminal olefin increases the
to 2 equiv., an excellent yield with 83% branched selectivity and energy barrier for branch-selective carbonylation and thus decreases
96% 3aa-2/iso selectivity was observed (Supplementary Table 1, the Markovnikov selectivity. For example, with 4-methylpentene as
entries 4 and 5).
the substrate, the branched selectivity drops slightly to 78% (isolated
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
3
© 2016 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.2586
ARTICLES
Table 2 | Markovnikov alkoxycarbonylation of propene and butenes.
X mol% PdBr2,
Y mol% L16,
Z mol% TsOH·H2O
O
O
R2
OR1
R1OH
Alkene
R2
+
+
OR1
R2 = Me or H
Toluene,
CO (40 bar), 20 h
2d R1 = PhCH2
2q R1 = PhCH2CH2
Branched
Linear
Entry*
Alkene
Propene
R1OH
X/Y/Z
Temperature (°C)
Yield (%)†
Branched selectivity (%)‡
TON
1
1o
1o
1o
1o
1o
1p
1q
1r
2d
2d
2d
2d
2d
2q
2q
2q
0.5/2/3
100
100
100
120
80
100
100
100
>99
>99
96
76
>99
96
88
86
84
81
89
81
200
2
3
4
5
6
7
8
Propene
Propene
Propene
Propene
1-butene
2-butene
Raffinate-1||
0.1/0.4/0.6
0.05/0.2/0.3
0.01/0.06/0.06
0.5/2/3
0.5/2/3
0.5/2/3
1,000
1,920
7,600§
200
192
180
90
79
81
81
0.5/2/3
158
*The Supplementary Information gives detailed procedures. †Isolated yield based on alcohol. ‡Selectivity determined by 1H NMR. §35 h. ||Raffinate-1: mixture of isobutene 42%, 1-butene 26%, 2-butene 17%,
1,3-butadiene 0.3% and butane/isobutane 15%. TsOH·H2O, p-toluenesulfonic acid monohydrate; TON, turnover number.
yield 78% (Table 1, entry 12)). Using 3-methylpentene, the steric concentration of 1-butene (Supplementary Fig. 2). An industrial
hindrance greatly increased; nevertheless, 49% branched selectivity mixture, Raffinate-1 (mixture of isobutene 42%, 1-butene 26%,
could be observed (d.r. = 1:1) (Table 1, entry 13). This was also ver- 2-butene 17%, 1,3-butadiene 0.3%, butane/isobutane 15%
ified by using β-citronellene as the substrate, in which a mixture of (Table 2)), also proved to be able to afford the corresponding
branched and linear products (52:48) was isolated with 88% yield ester with a 79% yield and 81% branched selectivity, with tert-
(Table 1, entry 14). (The trisubstituted C=C moiety of β-citronellene butoxyphenyl ethyl ether as the major by-product (Table 2, entry 8).
is unreactive under our reaction conditions.)
For various industrial applications, diverse aliphatic carboxylic
In addition to ethene, propene and butene are the most important acid esters with different alcohols are interesting. Hence, several alco-
industrial olefin feedstocks, which today are mainly produced by the hols were tested in the alkoxycarbonylation of 1-octene under the
classic cracking processes. However, emerging technologies enable optimized conditions (Table 3). Primary alcohols, such as ethanol,
the ‘on-purpose’ propene production from naturally occurring benzyl alcohol and nonanol, gave excellent yields (88–99%) and
sources, such as shale gas and even biomass (http://plasticsengine- high branched selectivities (83–85% (Table 3, entries 1, 3 and 5)).
eringblog.com/2013/02/20/how-shale-gas-is-changing-propylene/). Secondary alcohols, such as isopropanol and cyclohexanol, gave
Also, butenes are produced on an industrial scale by ethene dimeri- similar good results to those of primary alcohols (Table 3, entries 2
zation. New selective valorizations of these essential feedstocks are and 9). Similarly, using more-bulky substrates, for example,
highly interesting for the chemical industry.
menthol, led to good yield and selectivity (Table 3, entry 14).
Clearly, single-branched products could be obtained by selective However, the tertiary alcohol t-amyl-OH gave a yield of only 20%
alkoxycarbonylation of propene or butene. However, this apparently with a high branched selectivity (Table 3, entry 4). Apparently, the
simple transformation is not realized with an acceptable perform- steric bulkiness of the alcohol has no influence on the regioselectivity
ance23. To improve the reactivity, we performed the catalytic tests of the reaction, but influences the reactivity. From a synthetic point
in the presence of 3 mol% PTSA. We assumed that the additional of view, it is interesting that heteroaromatic alcohols were also toler-
acid will act as a more-efficient reaction promoter compared with ated. In fact, furyl- or thiophenyl-based alcohols gave ∼80%
water, and did not worry about the formation of other internal branched selectivity (Table 3, entries 6 and 7). However, applying
isomeric esters. Indeed, in the presence of PTSA and PdBr2 as a less-basic alcohols, namely phenol, pentafluorobenzyl alcohol and
better palladium source, we obtained improved results. With propene CF3CH2OH, gave a decreased branched selectivity (65, 72 and
and benzyl alcohol as the starting materials, benzyl isobutyrate (see 65%, respectively (Table 3, entries 8, 12 and 13)). In the case of
Table 2) was obtained in quantitative yield with a branched selectiv- (4-bromophenyl)methanol, no activation of the halide was observed
ity of 88% (Table 2, entry 1). The catalyst loading could be decreased and high yield and branched selectivity were obtained (Table 3, entry
further to 0.05 mol% with almost quantitative yield and a slight 11). The more-demanding menthol gave 82% yield and 83%
decrease of branched selectivity (96% yield, 84% branched selectiv- branched selectivity (Table 3, entry 14), whereas with 1-phenyletha-
ity (Table 2, entry 3)). A further decrease of the catalyst loading to nol as the substrate, a d.r. 1:1 mixture of the product was obtained
0.01 mol% gave 76% yield with 81% branched selectivity at 120 °C with 89% yield and 85% branched selectivity (Table 3, entry 15).
(Table 2, entry 4). Moreover, a decrease of the reaction temperature
to 80 °C also gave quantitative yield at a catalyst loading of 0.5%, Computational studies. To understand this observed uncommon
with 89% branched selectivity (Table 2, entry 5).
regioselectivity in favour of the branched products, we carried out
For the study of butenes, 2-phenylethanol was taken as a model computation using the range-separated hybrid WB97XD
substrate given that the expected branched product phenethyl functional, which includes dispersion correction47,48. In our study,
2-methylbutyrate is used as a floral and fruity organoleptic to we used the cationic [L2PdH]+ complex as the active catalyst and
provide flavour and fragrance in industry. With 1-butene as the sub- propene as the model olefin as well as the computed Gibbs free
strate, a yield of 96% was obtained, with 81% selectivity of the energies, which included the solvation effect of toluene, to discuss
branched product. More strikingly, with 2-butene, a similar yield the regioselectivity. To differentiate the ligand effect, calculations
and selectivity (90% yield, 81% branched selectivity (Table 2)) was were performed with L1 (triphenylphosphine) and L15 (1-(2-
also obtained. By comparing the reaction kinetic profiles of methoxyphenyl)-1H-pyrrolyl diphenylphosphine). To enable a direct
1-butene and 2-butene from gas-consumption experiments, we can and systematic comparison between computation and experiment,
clearly see that the reaction rate of 1-butene at the initial stage is we carried out additional catalytic experiments using both L1 and
faster than that of 2-butene, probably because of the higher L15 as ligands and propene as the substrate (Supplementary Table 2).
4
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
© 2016 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.2586
ARTICLES
Table 3 | Alcohol substrate scope of Markovnikov alkoxycarbonylation of 1-octene.
1 mol% PdCl2,
4 mol% L16
3ax-1: 1-COOR
3ax-2: 2-COOR
3ax-3: 3-COOR
3ax-4: 4-COOR
4
2
COOR
C4H9
+
ROH
C4H9
3
1
Toluene, 100 °C, 20 h,
CO (40 bar), 5µL H2O
1a
R–OH
2x
Entry*
Yield (%)†
Branched selectivity (%)‡
3ax-2/iso (%)‡
1
EtOH
i-PrOH
BnOH
t-Amyl-OH
Nonanol
2b
88
82
99
20
99
85
85
83
82
84
97
2
3
4
5
2c
2d
2e
2f
98
98
>99
98
O
6
OH
2g
90
80
>99
S
OH
7
2h
91
80
>99
8
9
Phenol
2i
2j
74
96
65
86
98
96
OH
10
11
2k
2l
94
95
85
84
98
96
MeO
Br
OH
OH
F
F
F
F
12
2m
83
72
>99
OH
F
F3C OH
13
14
2n
2o
63
82
65
83
99
96
HO
15
2p
89
85
98
Ph OH
*Reaction conditions: 1 (2 mmol), 2 (1 mmol), PdCl2 (1 mol%), L16 (4 mol%), CO (40 bar), toluene, 5 µl H2O, 100 °C, 20 h. †Isolated yield based on alcohol. ‡Branched selectivity and 3ax-2/iso ratios were
determined by GC.
On the basis of the proposed reaction mechanism for the cationic
For L15 (Fig. 3a), [(L15)2Pd(isopropyl)]+ is more stable than
[L2PdH]+ complexes (Supplementary Fig. 4)41, two elementary steps [(L15)2Pd(propyl)]+ by 2.38 kcal mol–1, and the thermodynamically
might determine the regioselectivity. The first one is the Pd–H bond more-stable [(L15)2Pd(isopropyl)]+ should lead to the branched
insertion into the C=C double bond of the coordinated olefin with product. However, the expected regioselectivity (96/4) is much
the formation of the linear and branched alkyl complexes. The higher than that in the catalytic experiment (75/25). For L1,
second possibility is CO coordination to the alkyl complex and [(L1)2Pd(isopropyl)]+ is more stable than [(L1)2Pd(propyl)]+ by
subsequent CO insertion or carbonylation. Both options were 0.78 kcal mol–1, and the expected regioselectivity (74/26) is also
computed for comparison (Fig. 3).
With both L1 and L15 asthe ligands, we found stable complexes for that Pd–H insertion does not determine the regioselectivity.
the coordinated propene [L2PdH(propene)]+, in which the C=C
Next, we computed the reaction energy of CO coordination
much higher than that in the experiment (59/41). This indicates
double bond is roughly perpendicular to the Pd–H bond. However, and the relative stability of the [(L15)2Pd(CO)(propyl)]+ and
the corresponding transition states for the Pd–H insertion into the [(L15)2Pd(CO)(isopropyl)]+ complexes. It was found that the for-
C=C bond by rotating the C=C bond syn to the Pd–H bond could mation of [(L15)2Pd(CO)(propyl)]+ and [(L15)2Pd(CO)(isopropyl)]+
not be located, which indicates that the barrier is negligibly low. is endergonic by 1.23 and 2.79 kcal mol–1, respectively. On the
Therefore, we used the relative energies of the most-stable propyl basis of their energy difference (0.82 kcal mol–1), the ratio of
[L2Pd(propyl)]+ and isopropyl [L2Pd(isopropyl)]+ complexes for dis- [(L15)2Pd(CO)(isopropyl)]+ and [(L15)2Pd(CO)(propyl)]+ under equili-
cussion under the consideration that all stable [L2Pd(H)(propene)]+ brium is 75/25; such a ratio is equal to the observed regioselectivity
complexes can form the corresponding propyl [L2Pd(propyl)]+ and in favour of the branched product (Supplementary Table 2). Starting
isopropyl [L2Pd(isopropyl)]+ complexes.
from [(L15)2Pd(CO)(propyl)]+, CO insertion has a barrier of
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
5
© 2016 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.2586
ARTICLES
a
Using L15 as ligand
12.13 (11.66)
12.53 (12.01)
Pd
[(L15)2Pd(TS-CO-propyl)]+
[(L15)2Pd(TS-CO-isopropyl)]+
H
Pd
Pd
From the most stable
[(L15)2PdH(propene)]+
OC
OC
11.19 (10.88)
CO
Pd
CO
Pd
Pd
H
[(L15)2Pd(CO)(propyl)]+
Pd
H
[(L15)2Pd(CO)(isopropyl)]+
3.61 (2.87)
[(L15)2Pd(propyl)]+
2.79 (2.19)
[(L15)2Pd(isopropyl)]+
2.38 (2.14)
Pd
OC
0.00 (0.00)
Pd
OC
[(L15)2Pd(-CO-isopropyl)]+
0.82 (0.68)
–7.82 (–7.23)
[(L15)2Pd(-CO-propyl)]+
–8.06 (–7.37)
b
Using L1 as ligand
Pd
H
11.10 (11.02)
[(L1)2Pd(TS-CO-propyl)]+
11.02 (10.60)
[(L1)2Pd(TS-CO-isopropyl)]+
From the most stable
Pd
Pd
[(L1)2PdH(propene)]+
OC
OC
7.79 (7.49)
CO
Pd
CO
Pd
Pd
H
Pd
H
[(L1)2Pd(CO)(propyl)]+
[(L1)2Pd(CO)(isopropyl)]+
[(L1)2Pd(propyl)]+
Pd
OC
[(L1)2Pd(isopropyl)]+
2.02 (1.42)
1.85 (1.28)
0.78 (0.70)
0.00 (0.00)
Pd
OC
[(L1)2Pd(-CO-isopropyl)]+
[(L1)2Pd(-CO-propyl)]+
–8.46 (–8.12)
0.17 (0.14)
–7.74 (–7.13)
Figure 3 | DFT calculations. Gibbs free energies of Pd–H insertion and carbonylation for reactions using L15 and L1 on WB97XD-SCRF/TZVP//WB97XD/
SVP, which included the solvation of toluene (the WB97XD/TZVP//WB97XD/SVP gas-phase values are in parenthesis). a, DFT calculation based on L15.
b, DFT calculation based on L1. TS, transition state.
8.92 kcal mol–1 and is exergonic by 11.67 kcal mol–1, and starting difference in thermodynamic stability of [(L1)2Pd(CO)(isopropyl)]+
from [(L15)2Pd(CO)(isopropyl)]+, CO insertion has a barrier of and [(L1)2Pd(CO)(propyl)]+ should determine the regioselectivity.
9.34 kcal mol–1 and is exergonic by 11.67 kcal mol–1.
All these results clearly show that the enhanced stability of the iso-
As the back reaction of CO insertion has a higher barrier (20.59 propyl complexes [(L15)2Pd(CO)(isopropyl)]+ and [(L1)2Pd(CO)(iso-
and 19.95 kcal mol–1 for propyl and isopropyl acyl complexes, propyl)]+ should determine the regioselectivity of the overall process.
respectively), the difference in thermodynamic stability of Furthermore, the reaction with L15 is more regioselective than that
[(L15)2Pd(CO)(isopropyl)]+ and [(L15)2Pd(CO)(propyl)]+ should with L1, which is in agreement with our experimental findings.
determine the regioselectivity.
Interestingly, as shown in Fig. 3, the stabilities of [(L15)2Pd(CO)
For L1 (Fig. 3b), the formation of [(L1)2Pd(CO)(propyl)]+ and (isopropyl)]+ and [(L1)2Pd(CO)(isopropyl)]+ can be enhanced with
[(L1)2Pd(CO)(isopropyl)]+ is endergonic by 1.24 and 1.85 kcal toluene as the solvent compared with the gas phase. Inspired by
mol–1, respectively. The expected ratio of [(L1)2Pd(CO)(iso- these results, we tested the solvent effect on the regioselectivity,
propyl)]+ and [(L1)2Pd(CO)(propyl)]+ under equilibrium is 56/44. both experimentally and computationally. Indeed, using THF as
Such a ratio is close to the observed regioselectivity (59/41) in solvent, the experimentally observed regioselectivity increased
favour of the branched product (Supplementary Table 2). Starting from 75/25 to 90/10 for L15, as well as from 59/41 to 77/23 for
from [(L1)2Pd(CO)(propyl)]+, CO insertion has a barrier of L1. Computationally, the regioselectivity for L15 increases from
9.08 kcal mol–1 and is exergonic by 9.76 kcal mol–1, whereas starting 75/25 to 83/17; however, no such effect was found for L1.
from [(L1)2Pd(CO)(isopropyl)]+, CO insertion has a barrier of Additionally, we also found that the same regioselectivity (91%)
9.17 kcal mol–1 and is exergonic by 10.31 kcal mol–1.
for L15 in THF can be achieved at a lower temperature (70 °C)
As the back reaction of CO insertion (or the decarbonylation step) but in a lower yield (92 versus 60%). Furthermore, increased CO
has a higher barrier (18.84 and 19.48 kcal mol–1, respectively), the pressure affects neither the yield nor regioselectivity.
6
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
© 2016 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.2586
ARTICLES
21. Noonan, G. M., Fuentes, J. A., Cobley, C. J. & Clarke, M. L. An asymmetric
hydroformylation catalyst that delivers branched aldehydes from alkyl alkenes.
Angew. Chem. Int. Ed. 51, 2477–2480 (2012).
22. Klosin, J. & Landis, C. R. Ligands for practical rhodium-catalyzed asymmetric
hydroformylation. Acc. Chem. Res. 40, 1251–1259 (2007).
23. Cavinato, G. & Toniolo, L. Metal in organic synthesis. VI. The solvent effect on
the hydrocarbonalkoxylation of propene promoted by a [PdCl2(PPh3)2]−PPh3
catalyst precursor. J. Mol. Catal. 10, 161–170 (1981).
24. Rucklidge, A. J., Morris, G. E. & Cole-Hamilton, D. J. Methoxycarbonylation of
vinyl acetate catalysed by palladium complexes of bis
(ditertiarybutylphosphinomethyl)benzene and related ligands. Chem. Commun.
1176–1178 (2005).
25. Shuklov, I. A. et al. Promoters for Pd-catalyzed methoxycarbonylation of vinyl
acetate. Arkivoc 66–75 (2012).
26. Frew, J. J. R., Clarke, M. L., Mayer, U., Van Rensburg, H. & Tooze, R. P.
Palladium complexes of new bulky fluorinated diphosphines give particularly
active and regioselective catalysts for hydroxycarbonylation of styrene. Dalton
Trans. 1976–1978 (2008).
27. Fuentes, J. A., Slawin, A. M. Z. & Clarke, M. L. Application of palladium (trioxo-
adamantyl cage phosphine)chloride complexes as catalysts for the
alkoxycarbonylation of styrene; Pd catalysed tert-butoxycarbonylation of
styrene. Catal. Sci. Technol. 2, 715–718 (2012).
Conclusion
In conclusion, we have discovered a PdX2/N-phenylpyrrole
phosphine-type catalyst system for the regioselective Markovnikov
alkoxycarbonylation reactions of alkenes. Industrially important
aliphatic olefins without further functional groups can be alkoxy-
carbonylated selectively to the branched products. The general
applicability is demonstrated in 17 examples (71–99% yield, 49–91%
branched selectivity). Furthermore, diverse alcohols are also well
tolerated (15 examples, 20–99% yield, 65–85% branched selectivity).
The applicability of this methodology is highlighted further by the
employment of an industrial C4 mixture in the synthesis of phenethyl
2-methylbutyrate, a fruity fragrance. On the basis of density func-
tional theory (DFT) computation, including dispersion correction,
the experimentally observed branched selectivity should be deter-
mined by the enhanced thermodynamic stability of the corresponding
isopropyl [L2Pd(CO)(isopropyl)]+ complex over the propyl
[L2Pd(CO)(propyl)]+ complex. Attempts with further ligand
modifications are currently underway in our research group.
28. Grabulosa, A., Frew, J. J. R., Fuentes, J. A., Slawin, A. M. Z. & Clarke, M. L.
Palladium complexes of bulky ortho-trifluoromethylphenyl-substituted
phosphines: unusually regioselective catalysts for the hydroxycarbonylation and
alkoxycarbonylation of alkenes. J. Mol. Catal. A 330, 18–25 (2010).
29. Ooka, H., Inoue, T., Itsuno, S. & Tanaka, M. Highly active and selective
palladium catalyst for hydroesterification of styrene and vinyl acetate promoted
by polymeric sulfonic acids. Chem. Commun. 1173–1175 (2005).
30. Liu, J., Liu, Q., Franke, R., Jackstell, R. & Beller, M. Ligand-controlled
palladium-catalyzed alkoxycarbonylation of allenes: regioselective synthesis
of α,β- and β,γ-unsaturated esters. J. Am. Chem. Soc. 137, 8556–8563 (2015).
31. Konrad, T. M., Durrani, J. T., Cobley, C. J. & Clarke, M. L. Simultaneous control
of regioselectivity and enantioselectivity in the hydroxycarbonylation and
methoxycarbonylation of vinyl arenes. Chem. Commun. 49, 3306–3308 (2013).
32. Guiu, E. et al. Electronic effect of diphosphines on the regioselectivity of the
palladium-catalyzed hydroesterification of styrene. Organometallics 25,
3102–3104 (2006).
33. Lee, C. W. & Alper, H. Hydroesterification of olefins catalyzed by Pd(OAc)2
immobilized on montmorillonite. J. Org. Chem. 60, 250–252 (1995).
34. Alper, H., Woell, J. B., Despeyroux, B. & Smith, D. J. H. The regiospecific
palladium catalysed hydrocarboxylation of alkenes under mild conditions.
Chem. Commun. 1270–1271 (1983).
35. Williams, D. B. G., Shaw, M. L., Green, M. J. & Holzapfel, C. W. Aluminum
triflate as a highly active and efficient nonprotic cocatalyst in the
palladium-catalyzed methoxycarbonylation reaction. Angew. Chem. Int. Ed. 47,
560–563 (2008).
36. Jørgensen, M., Lee, S., Liu, X., Wolkowski, J. P. & Hartwig, J. F. Efficient synthesis
of α-aryl esters by room-temperature palladium-catalyzed coupling of aryl
halides with ester enolates. J. Am. Chem. Soc. 124, 12557–12565 (2002).
37. Wender, P. A., Koehler, M. F. T. & Sendzik, M. A new synthetic approach
to the C ring of known as well as novel bryostatin analogues. Org. Lett. 5,
4549–4552 (2003).
38. Crimmins, M. T., Carroll, C. A. & Wells, A. J. Pinacol-type rearrangements of
intramolecular photocycloadducts: application of the 2,2-dimethyl-4-pentenoate
protecting group. Tetrahedron Lett. 39, 7005–7008 (1998).
39. Gauthier, D., Lindhardt, A. T., Olsen, E. P. K., Overgaard, J. & Skrydstrup, T.
In situ generated bulky palladium hydride complexes as catalysts for the efficient
isomerization of olefins. Selective transformation of terminal alkenes to
2-alkenes. J. Am. Chem. Soc. 132, 7998–8009 (2010).
40. Larionov, E., Lin, L., Guénée, L. & Mazet, C. Scope and mechanism in
palladium-catalyzed isomerizations of highly substituted allylic, homoallylic,
and alkenyl alcohols. J. Am. Chem. Soc. 136, 16882–16894 (2014).
41. Grushin, V. V. Hydrido complexes of palladium. Chem. Rev. 96,
2011–2034 (1996).
Received 12 November 2015; accepted 30 June 2016;
published online 5 September 2016
References
1. Trost, B. M. Atom economy—a challenge for organic synthesis:
homogeneous catalysis leads the way. Angew. Chem. Int. Ed. Engl. 34,
259–281 (1995).
2. Markownikoff, W. I. Ueber die Abhängigkeit der verschiedenen Vertretbarkeit
des Radicalwasserstoffs in den isomeren Buttersäuren. Justus Liebigs Ann. Chem.
153, 228–259 (1870).
3. Huang, L., Arndt, M., Gooßen, K., Heydt, H. & Gooßen, L. J. Late transition
metal-catalyzed hydroamination and hydroamidation. Chem. Rev. 115,
2596–2697 (2015).
4. Müller, T. E., Hultzsch, K. C., Yus, M., Foubelo, F. & Tada, M. Hydroamination:
direct addition of amines to alkenes and alkynes. Chem. Rev. 108,
3795–3892 (2008).
5. Whittlesey, M. K. in Comprehensive Coordination Chemistry II (eds McCleverty,
J. A. & Meyer, T. J.) 265–304 (Pergamon, 2003).
6. Müller, T. E. & Beller, M. Metal-initiated amination of alkenes and alkynes.
Chem. Rev. 98, 675–704 (1998).
7. Beller, M., Seayad, J., Tillack, A. & Jiao, H. Catalytic Markovnikov and anti-
Markovnikov functionalization of alkenes and alkynes: recent developments and
trends. Angew. Chem. Int. Ed. 43, 3368–3398 (2004).
8. Mahatthananchai, J., Dumas, A. M. & Bode, J. W. Catalytic selective synthesis.
Angew. Chem. Int. Ed. 51, 10954–10990 (2012).
9. Weissermel, K. & Arpe, H.-J. Industrial Organic Chemistry 127–144
(Wiley-VCH, 2008).
10. Kégl, T. in Modern Carbonylation Methods (ed Kollár, L.) 161–198
(Wiley-VCH, 2008).
11. Cornils, B. & Hermann, W. A. Applied Homogeneous Catalysis with
Organometallic Compounds (Wiley-VCH, 2008).
12. El Ali, B. & Alper, H. in Handbook of Organopalladium Chemistry for Organic
Synthesis (ed Negishi, E.-I.) 2333–2349 (John Wiley & Sons, 2003).
13. Franke, R., Selent, D. & Börner, A. Applied hydroformylation. Chem. Rev. 112,
5675–5732 (2012).
14. Ali, B. E. & Alper, H. in Transition Metals for Organic Synthesis (eds Beller, M. &
Bolm, C.) 113–132 (Wiley-VCH, 2008).
15. Beller, M., Cornils, B., Frohning, C. D. & Kohlpaintner, C. W. Progress in
hydroformylation and carbonylation. J. Mol. Catal. A 104, 17–85 (1995).
16. Jimenez Rodriguez, C., Foster, D. F., Eastham, G. R. & Cole-Hamilton, D. J.
Highly selective formation of linear esters from terminal and internal alkenes
catalysed by palladium complexes of bis-(di-tert-butylphosphinomethyl)
benzene. Chem. Commun. 1720–1721 (2004).
17. Kiss, G. Palladium-catalyzed Reppe carbonylation. Chem. Rev. 101,
3435–3456 (2001).
18. Clegg, W. et al. Highly active and selective catalysts for the production of methyl
propanoate via the methoxycarbonylation of ethene. Chem. Commun.
1877–1878 (1999).
42. Negishi, E.-I. in Handbook of Organopalladium Chemistry for Organic Synthesis
(ed Negishi, E.-I.) 2783–2788 (John Wiley & Sons, 2003).
43. Elali, B. & Alper, H. Formic-acid palladium acetate-1,4-bis(diphenylphosphino)
butane—an effective catalytic system for regioselective hydrocarboxylation of
simple and functionalized olefins. J. Mol. Catal. 77, 7–13 (1992).
44. Pugh, R. I., Drent, E. & Pringle, P. G. Tandem isomerisation–carbonylation
catalysis: highly active palladium(II) catalysts for the selective
19. Jamie, T. & Durrani, M. L. C. in Stereoselective Synthesis of Drugs and Natural
Products (eds Andrushko, N. & Andrushko, V.) Part 2, Ch. 14 (John Wiley &
Sons, 2013).
methoxycarbonylation of internal alkenes to linear esters. Chem. Commun.
1476–1477 (2001).
45. Surry, D. S. & Buchwald, S. L. Dialkylbiaryl phosphines in Pd-catalyzed
amination: a user’s guide. Chem. Sci. 2, 27–50 (2011).
46. Zapf, A. et al. Practical synthesis of new and highly efficient ligands for the
Suzuki reaction of aryl chlorides. Chem. Commun. 38–39 (2004).
20. Noonan, G. M., Cobley, C. J., Mahoney, T. & Clarke, M. L. Rhodium/
phospholane–phosphite catalysts give unusually high regioselectivity in the
enantioselective hydroformylation of vinyl arenes. Chem. Commun. 50,
1475–1477 (2014).
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
7
© 2016 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.2586
ARTICLES
47. Chai, J.-D. & Head-Gordon, M. Long-range corrected hybrid density functionals
with damped atom–atom dispersion corrections. Phys. Chem. Chem. Phys. 10,
6615–6620 (2008).
48. Kumar, M., Chaudhari, R. V., Subramaniam, B. & Jackson, T. A. Ligand effects
on the regioselectivity of rhodium-catalyzed hydroformylation: density
functional calculations illuminate the role of long-range noncovalent
interactions. Organometallics 33, 4183–4191 (2014).
Author contributions
M.B. and H.L. conceived and designed the experiments. H.L. and K.D. performed the
experiments and analysed the data. H.J. performed the DFT study. H.N. and R.J.
participated in the discussions and supported the project. M.B., H.J. and H.L.
co-wrote the paper.
Additional information
Supplementary information and chemical compound information are available in the
online version of the paper. Reprints and permissions information is available online
at www.nature.com/reprints. Correspondence and requests for materials should be
addressed to M.B.
Acknowledgements
The Federal Ministry of Education and Research (BMBF) and Evonik Industries are
acknowledged for their general support. K.D. acknowledges a Shanghai Institute of
Organic Chemistry-Zhejiang Medicine joint fellowship for financial support. The
analytic department of LIKAT is acknowledged. We are thankful to Q. Liu and
X. Fang for the helpful and supportive discussions.
Competing financial interests
The authors declare no competing financial interests.
8
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
© 2016 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.