High linear regioselectivity in the rhodium-catalyzed hydro(deuterio) formylation of 3,4,4-trimethylpent-1-ene: The role of β-hydride elimination

Article abstract of DOI:10.1016/j.molcata.2011.12.021

The regioselectivity in the hydroformylation reaction catalyzed by an unmodified Rh catalyst has been investigated for a number of α-methylsubstituted alk-1-enes (3-methylbut-1-ene MB₁, 3-methylpent-1-ene MP₁, 3,4-dimethylpent-1-ene DMP₁, and 3,4,4-trimethylpent-1-ene TMP₁) experimentally (at 20 °C and 100 atm CO/H₂ total pressure) and theoretically at the B3P86/6-31G* level with Rh described by effective core potentials in the LanL2DZ valence basis set. For all substrates the formation of the linear aldehyde (L) with respect to the branched one (B) in a prevailing amount has been observed (L/B > 62/38); the L isomer was formed as the almost exclusive product in the case of TMP₁ (L/B = 95/5). ²H NMR investigations of crude reaction mixtures, coming from analogous deuterioformylation experiments interrupted at partial substrate conversion, showed that in the case of TMP ₁ only the branched alkyl-rhodium intermediate, precursor of the branched aldehyde, via β-hydride elimination mainly generates terminal deuterated olefins and, to a lesser extent, internal ones. The reversibility of the branched alkyl-Rh intermediates accounts for the high regioselectivity in favor of the linear aldehyde. Computational studies confirm the importance of the alkyl-Rh transition state (TS) stability to reproduce the experimental regioselectivity, or even to predict it, when the reaction is nonreversible (i.e. for MB₁, MP₁, and DMP₁). In the case of TMP₁, the free energy profiles for further reaction steps along branched and linear pathways have been examined to elucidate the origin of reaction reversibility. The TS for the alkyl migratory insertion onto the CO coordinated to rhodium, higher than that for the alkyl-Rh intermediate formation, explains the reason why in deuterioformylation experiments at partial conversion the monodeuterated terminal olefin TMP₁-1-d₁ is obtained. This occurs for one out of two most populated reactant conformers of TMP₁, although for the Curtin-Hammett principle reactant populations are not particularly important. For the other, the reaction proceeds to the branched aldehyde. Only for a less populated reactant conformer the internal olefin is obtained. Conversely, along the linear pathway the CO addition and alkyl migratory insertion steps occur, respectively, in a practically spontaneous way and with very low TS in any case. Agostic interactions (using the QTAIM theory) and kinetic isotope effects have been evaluated and discussed. The examination of further reaction steps for DMP ₁ allowed us to demonstrate that the reaction is nonreversible for that substrate, despite the similarity between DMP₁ and TMP ₁. The tert-butyl group exerts its steric hindrance mainly on the very first branched reaction steps, favoring an alkyl-Rh TS arrangement lower in free energy than the alkyl-Rh migratory insertion onto the coordinated CO. In part the branched material returns to the reactant complex, thus enriching the linear fraction.

Full text of DOI:10.1016/j.molcata.2011.12.021

Journal of Molecular Catalysis A: Chemical 356 (2012) 1–13
Contents lists available at SciVerse ScienceDirect
Journal of Molecular Catalysis A: Chemical
journal homepage: www.elsevier.com/locate/molcata
High linear regioselectivity in the rhodium-catalyzed hydro(deuterio)formylation
of 3,4,4-trimethylpent-1-ene: The role of ␤-hydride elimination
Raffaello Lazzaronia,∗, Roberta Settambolob, Giuliano Alagonac,∗∗, Caterina Ghioc
a DCCI, University of Pisa, Via Risorgimento 35, I-56126 Pisa, Italy
b CNR-ICCOM, Pisa UOS, Via Risorgimento 35, I-56126 Pisa, Italy
c CNR-IPCF, Pisa UOS, Molecular Modeling Lab, Via Moruzzi 1, I-56124 Pisa, Italy
a r t i c l e i n f o
a b s t r a c t
Article history:
The regioselectivity in the hydroformylation reaction catalyzed by an unmodified Rh catalyst has been
investigated for a number of ␣-methylsubstituted alk-1-enes (3-methylbut-1-ene MB1, 3-methylpent-1-
ene MP1, 3,4-dimethylpent-1-ene DMP1, and 3,4,4-trimethylpent-1-ene TMP1) experimentally (at 20 ◦C
and 100 atm CO/H2 total pressure) and theoretically at the B3P86/6-31G* level with Rh described by effec-
tive core potentials in the LanL2DZ valence basis set. For all substrates the formation of the linear aldehyde
(L) with respect to the branched one (B) in a prevailing amount has been observed (L/B > 62/38); the L iso-
mer was formed as the almost exclusive product in the case of TMP1 (L/B = 95/5). 2H NMR investigations of
crude reaction mixtures, coming from analogous deuterioformylation experiments interrupted at partial
substrate conversion, showed that in the case of TMP1 only the branched alkyl-rhodium intermediate,
precursor of the branched aldehyde, via ␤-hydride elimination mainly generates terminal deuterated
olefins and, to a lesser extent, internal ones. The reversibility of the branched alkyl-Rh intermediates
accounts for the high regioselectivity in favor of the linear aldehyde. Computational studies confirm the
importance of the alkyl-Rh transition state (TS) stability to reproduce the experimental regioselectiv-
ity, or even to predict it, when the reaction is nonreversible (i.e. for MB1, MP1, and DMP1). In the case
of TMP1, the free energy profiles for further reaction steps along branched and linear pathways have
been examined to elucidate the origin of reaction reversibility. The TS for the alkyl migratory insertion
onto the CO coordinated to rhodium, higher than that for the alkyl-Rh intermediate formation, explains
the reason why in deuterioformylation experiments at partial conversion the monodeuterated terminal
olefin TMP1-1-d1 is obtained. This occurs for one out of two most populated reactant conformers of TMP1,
although for the Curtin–Hammett principle reactant populations are not particularly important. For the
other, the reaction proceeds to the branched aldehyde. Only for a less populated reactant conformer
the internal olefin is obtained. Conversely, along the linear pathway the CO addition and alkyl migra-
tory insertion steps occur, respectively, in a practically spontaneous way and with very low TS in any
case. Agostic interactions (using the QTAIM theory) and kinetic isotope effects have been evaluated and
discussed. The examination of further reaction steps for DMP1 allowed us to demonstrate that the reac-
tion is nonreversible for that substrate, despite the similarity between DMP1 and TMP1. The tert-butyl
group exerts its steric hindrance mainly on the very first branched reaction steps, favoring an alkyl-Rh
TS arrangement lower in free energy than the alkyl-Rh migratory insertion onto the coordinated CO. In
part the branched material returns to the reactant complex, thus enriching the linear fraction.
© 2011 Elsevier B.V. All rights reserved.
Received 21 July 2011
Received in revised form
26 November 2011
Accepted 18 December 2011
Available online 27 December 2011
Keywords:
Hydroformylation
Deuterioformylation
DFT
B3P86/6-31G*
ECP/LanL2DZ
Regioselectivity
3-Methylbut-1-ene
3-Methylpent-1-ene
3,4-Dimethylpent-1-ene
3,4,4-Trimethylpent-1-ene
1. Introduction
alkyl-metal intermediates along the pathway to the linear, L, or
branched, B, product aldehydes (Scheme 1), can be a reversible or
nonreversible step, primarily depending on the temperature and
the substrate nature.
Under rhodium-catalyzed hydroformylation conditions, the
alkene insertion into the Rh H bond, which gives rise to the
We found that at 100 ◦C ␤-hydride elimination occurs for vari-
ous vinyl substrates such as ethyl vinyl ether, 1-hexene, allyl ethyl
ether, styrene and other aromatic substrates. While for aliphatic
vinyl [1] and allyl alkenes [1–3] ␤-hydride elimination involves
both the linear and branched alkyl intermediates, for the aro-
matic substrates (styrene) it occurs primarily for the branched ones
[4–9]. By contrast, at low temperature (<25 ◦C), these substrates do
∗
Corresponding author at: DCCI, University of Pisa, Via Risorgimento 35, I-56126
Pisa, Italy. Tel.: +39 0502219 227; fax: +39 0502219 260.
∗∗
Corresponding author at: CNR-IPCF, Pisa UOS, Molecular Modeling Lab, Via
Moruzzi 1, I-56124 Pisa, Italy. Tel.: +39 050315 2450; fax: +39 050315 2442.
E-mail addresses: lazza@dcci.unipi.it (R. Lazzaroni), G.Alagona@ipcf.cnr.it,
G.Alagona@pi.ipcf.cnr.it (G. Alagona).
1381-1169/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.molcata.2011.12.021

2
R. Lazzaroni et al. / Journal of Molecular Catalysis A: Chemical 356 (2012) 1–13
OHC
Z
H
branched
linear
Rh
Z
H
*
*
B
[Rh-H]
b
l
[Rh-H]
Z
Z
H
Z
CHO
H
Z
Rh
L
Scheme 1. Regioselectivity in the hydroformylation reaction of a terminal olefin. * Reversibility depending on the temperature and the substrate nature.
Table 1
Regioselectivity values for the hydroformylation of ␣-methylsubstituted alk-1-enes under 20 ◦C and 100 atm CO/H2 (1:1), at partial (20%) and complete substrate conversion.
R
CO/H2
R
R
O
+
O
Rh4(CO)12
CH3
T = 20°C
CH3
1
CH3
B
benzene
L
.
1
R
L:B(20% conv.)
L:B (complete conversion)
Experimental data
Experimental data
Computational resultsa
ꢀE
ꢀG
62:38
66:34
68:32
85:15
MB1
MP1
DMP1
TMP1
Me
Et
i-Pr
t-Bu
63:37
67:33
74:26
96:4
62:38
66:34
72:28
95:5
62:38
62:38
58:42
53:47
aB3P86/6-31G*/Lanl2DZ L:B ratios based on either potential (ꢀE) or free (ꢀG) energies.
not give ␤-hydride elimination and thus the alkyl formation is a
nonreversible step determining the regioselectivity of the process.
In order to establish whether the alkyl formation is a reversible
step at low temperature also for vinyl substrates bearing a bulky
substituent in ␣ position with respect to the double bond, some 3-
alkyl substituted alk-1-enes (see Table 1 for their definition) were
hydroformylated by us. Four different R moieties were taken into
account: they are in turn the methyl (Me), ethyl (Et), iso-propyl (iPr),
and tert-butyl (tBu) groups, leading respectively to 3-methylbut-
1-ene (MB1), 3-methylpent-1-ene (MP1), 3,4-dimethylpent-1-ene
(DMP1), and 3,4,4-trimethylpent-1-ene (TMP1).1
Deuterioformylation runs carried out under the same hydro-
formylation conditions, at partial substrate conversion, were
exploited to elucidate the chemical behavior of the 1-alkenes taken
into account. In particular, deuterioformylations at partial substrate
conversion of the substrate bring out significant information on the
possible reaction reversibility in the alkyl formation, as previously
remarked in the literature also by other authors for the hydro-
formylation of a variety of substrates employing different catalytic
precursors [10–12].
2. Results and discussion
2.1. Hydroformylation
Hydroformylation of the alk-1-enes (Table 1) was carried out
in benzene both at partial and complete substrate conversion in
a stainless-steel autoclave with Rh4(CO)12 as catalyst precursor, at
room temperature, under 100 atm total pressure (CO/H2 = 1:1). The
composition of the reaction mixtures was evaluated by GC analysis,
using n-decane as the internal standard.
In the case of the simplest olefins (MB1 and MP1) a slight L-
regioselectivity occurs, the L:B molar ratio varying from 62:38 to
66:34 both at partial and total substrate conversion. A larger preva-
lence of the L isomer (L:B = 72:28) was observed in the case of DMP1
bearing an iPr group at C3. The regioselectivity turns out to be
almost exclusively in favor of the L-regioisomer with TMP1, a sub-
strate characterized by the presence of a t-butyl group in ␣ position
with respect to the double bond.
2.2. Deuterioformylation
From a computational view-point, in the absence of ␤-hydride
elimination, the regioselectivity of the reaction can be evaluated on
the basis of the potential or free energy of the isomeric alkyl forma-
tion, provided all the possible conformers are taken into account,
and compared with the regioselectivity of aldehyde formation [13].
Conversely, when the reaction is reversible, the subsequent reac-
tion steps need to be examined with theoretical calculations at the
free energy level [14] (refer to the Computational Details section for
a description of employed methods) to investigate and eventually
rationalize the origin of reversibility. Kinetic isotope effects (KIE)
[15] have also been evaluated as well as the agostic interaction in
the alkyl-tricarbonyl intermediates in terms of the QTAIM theory
[16].
In order to establish if the Rh-alkyl formation is or not a
reversible step of the reaction, deuterioformylation experiments
with the above substrates were carried out. The same approach
had been previously employed with the same purpose by Casey
and Petrovich [10] and Nozaki et al. [12] among the authors here
cited. In the case of TMP1, the runs were carried out at both partial
and total conversion, at 20 ◦C under 100 atm of CO and D2 (1/1) at
constant pressure. The reaction was stopped after about 40 min at
20% substrate conversion into aldehydes. The reaction mixture was
also examined after 3 h at complete substrate conversion.
At the two selected conversion degrees (20% and total conver-
sion), samples of the reaction mixtures, containing unconverted
olefins and products, were directly analyzed without any han-
dling or treatment. The regioselectivity was high and, as in the
hydroformylation, in favor of the linear aldehyde. Inspection of the
2H NMR spectra allowed rapid and complete identification of all
deuterated species present in solution. The 2H NMR resonances
were assigned by comparison with the analogous signals present
1
Apart the first substrate (MB1), the others bear a chiral group. Both enantiomers
(indicated using a plain or a primed name) need to be computed to consider the
attack onto either face of the double bond as well as that the chiral substrates are
used in racemic form in the experiments.

R. Lazzaroni et al. / Journal of Molecular Catalysis A: Chemical 356 (2012) 1–13
3
Table 2
well as the isotopic effect that favors the Rh-H elimination with
respect to the Rh-D one [15c].
1H NMR chemical shifts (␦, ppm) of species arising from hydroformylation of TMP1
at partial substrate conversion.
As reported in Scheme 2, in the light of the generally accepted
mechanism of the hydroformylation reaction, the branched alkyl-
rhodium intermediate (b) gives, via ␤-hydride elimination, the
complex between the deuterated olefins TMP1-1-d1 and Rh-H.
Exchange of labeled olefins with unlabeled TMP1 gives the non-
deuterated ␩2H–TMP1 complex and the free deuterated terminal
olefins. The same alkyl can also undergo ␤-elimination to give the
internal olefin TMP2-1-d1. It is known that the ␤-hydride elimi-
nation is controlled by a kinetic isotope effect [15c], which means
that elimination of Rh-H is faster than that of Rh-D; this effect could
account for the observed accumulation of deuterated species in the
unconverted substrate.
Only in part the branched alkyl gives, via the normal hydro-
formylation steps, a small amount (5%) of the branched aldehyde.
Conversely, no ␤-elimination occurs for the linear alkyl (l) (no ter-
minal olefin deuterated at C2 of the double bond was observed),
while it completely evolves into the corresponding L aldehyde.
The ␩2H–TMP1 complex brings about the monodeuterated alde-
hyde without D in ␤-position (L-1-d1) (Scheme 2 and Table 3). This
accounts for the intensity of the CDO signal higher than that of the
D in ␤-position.
CH3
H
H
H
H
O
γ
CH3
ω
β
α
CH3 CH3
H
H
δ
.
Proton type
Signal
ı (ppm)
a
b
C6H6
CDCl3
␣
␥
␤
␦
2 Multiplets
Multiplet
Multiplet
Doublet
1.9 and 1.75
1.62
0.81
0.55
0.7
2.4 and 2.28
1.82
1.1
0.84
0.8
␻
Singlet
a
Reaction solvent.
After removal of benzene.
b
in the 1H NMR spectra (solvent C6D6) of the corresponding hydro-
formylation mixtures (the chemical shifts obtained are reported in
Table 2).
It is worth noting that the values in benzene as a solvent are
quite different from the values obtained in CDCl3. In particular, the
chemical shifts of the L isomer for the CH2 in ␤ position with respect
to the formyl group are 0.81 in benzene and 1.1 ppm in CDCl3.
In the 2H NMR spectrum relative to deuterioformylation experi-
ments at 20% conversion (Fig. 1 and Table 3), the two signals at 9.32
and 0.80 ppm are due to the deuterium atom of the formyl group
into L-1,3-d2 and L-1-d1 aldehydes and to the deuterium atom on
the carbon in ␤-position into L-1,3-d2 one, respectively. The inten-
sity of the formyl group signal is higher than that of the deuterium
in ␤-position with a ratio of 1:0.65 (vide infra for a discussion).
The sample also shows a significant resonance at 4.93 ppm
which can be assigned to a deuterium atom bonded to the termi-
nal carbon atom of the unconverted olefin TMP1-1-d1. In addition,
while the signal at 1.49 ppm is likely due to a deuterium atom
bonded to the terminal carbon atom of 3,4,4-trimethylpent-2-ene
(TMP2-1-d1), the signal at 1.602 ppm can be assigned to the deu-
terium atom CHDCDO in ␣ position to the carbonyl group arising
from deuterioformylation of the substrate monodeuterated in ter-
minal position TMP1-1-d1 (L-1,2,3-d3).
Deuterioformylation of DMP1 was also carried out under the
same experimental conditions adopted for TMP1. A regioselectiv-
ity ratio (L/B) of 72/28 was observed, such as a value similar to
that found for hydroformylation. 2H NMR spectrum of a sample of
the crude reaction mixture recovered at 20% substrate conversion
shows the presence of the signal due to CDO and the resonances
for CHD of the linear aldehyde and CH2D of the branched isomer
respectively. No traces of terminal or internal olefin deuterated in
terminal position have been observed. An accurate analysis of GC
and GC–MS showed the presence of a small amount (less than 5%)
of isomerized olefins.
It is noteworthy that for the simplest substrates, i.e. MB1 and
MP1, ␤-hydride elimination does not occur and only two signals
for the linear aldehyde and two signals for the branched one were
observed in 2H NMR spectra.
In order to explain the observed linear regioselectivity, the elec-
tronic effect of the R group could be taken in account, the +I
influence of the t-butyl group being stronger than those of i-Pr
and Me. Nevertheless, the steric hindrance is mainly responsi-
ble for the ␤-elimination. In particular the bulky t-butyl group in
TMP1, depending on its arrangement, can hamper the evolution of
the branched alkyl to the corresponding branched aldehyde and
favor ␤-hydride elimination. For the other branched alkyls, the
steric effect exerted by a sec-alkyl group bonded to the vinyl group
accounts for the prevalence of the linear isomer with respect to the
branched one, the steric hindrance being not enough to determine
␤-hydride elimination. Thus, with TMP1, unlike what observed for
all vinyl olefins previously investigated [1–3], ␤-hydride elimina-
tion at room temperature occurs, and exclusively for the branched
alkyl intermediate.
The GC–MS spectra, in addition to the unconverted substrate,
show the presence of isomerised olefins. No traces of hydrogena-
tion products were observed.
The formation of deuterated olefins under deuterioformylation
conditions can be reasonably explained taking into account the
reversible formation of the tertiary alkyl-metal intermediate as
2.3. Theoretical investigations on ˇ-hydride elimination
Theoretical regioselectivities computed under the hypothesis
of nonreversibility reproduce the experimental ratios at complete
conversion (Table 1) exactly for MB1 and almost exactly for MP1
when the potential energy-based values, also reported in Table 1
(fifth column), are used, as earlier found for other substrates [13]
as well. When Et is replaced with either an i-Pr (DMP1) or a t-
Bu (TMP1) group, theoretical ratio underestimates with respect to
the experimental values turn out to be increasingly more evident
(58:42 vs. 72:28 and 53:47 vs. 95:5) unlike for the other substrates.
Interestingly enough, when free energy-based values [14], also
Fig. 1. 2H NMR spectrum (46 MHz, 25 ◦C, C6D6 as external standard) of the crude
reaction mixture obtained by deuterioformylation of TMP1 at 20% substrate conver-
sion.

4
R. Lazzaroni et al. / Journal of Molecular Catalysis A: Chemical 356 (2012) 1–13
D
CDO
L-1,3-d2
TMP1
η2
Rh-H
D
D
Rh
D
-TMP1
l
H
TMP1-2-d1
Rh-D
+Rh-D
-Rh-D
TMP1
Rh-H
D
D
TMP1
η2
η2
-TMP1
-TMP1
H
D
TMP1-1-d1
TMP2-1-d1
Rh
D
TMP1
b
Rh-H
D
D
η2
-TMP1
H
CDO
D
B-1,3-d2
Rh-H
H
Rh
H
CDO
η2
-TMP1
l
L-1-d1
H
Scheme 2. Regioselectivity in the deuterioformylation reaction of TMP1.
Scheme 3. Mechanism of hydroformylation proposed by Wilkinson et al. [35] for a terminal olefin.

R. Lazzaroni et al. / Journal of Molecular Catalysis A: Chemical 356 (2012) 1–13
5
Table 3
2H NMR chemical shifts (ı, ppm)a of deuterated species arising from deuterioformylation of TMP1 at partial substrate conversion.
CDO
H
L-1-d1
D
CDO
D
CDO
D
CDO
D
SPECIES
D
D
TMP1-1-d1
4.93
TMP2-1-d1
1.49
L-1,3-d2
L-1,3-d2
L-1,2,3-d3
δ
0.80
9.32
1.60
aReferred to C6D6 as external standard; 46 MHz, C6H6, 25 ◦C.
reported in Table 1 (last column), are used, theoretical results for
the alkyl-Rh TS exactly reproduce the experimental ratios for MB1
and MP1, while for DMP1 the calculated ratio (68:32) turns out to
be in quite good agreement with the experiment. Actually, even for
TMP1 the regioselectivity evaluated employing free energy-based
results (L:B = 85:15) is satisfactory, being just slightly underesti-
mated with respect to the experimental value (L:B = 95:5). Anyway,
the experimental evidence of ␤-hydride elimination in the case
of TMP1 prompted us to theoretically investigate the subsequent
reaction steps (Scheme 3) to elucidate and possibly rationalize the
origin of reaction reversibility. To this end, all the stationary points
need to be located on the potential energy surface and their free
energies need to be computed because of the change in the num-
ber of species along the reaction pathways. Primarily the fourth CO
approaching path that played a critical role in 1,1-diphenylethene
[17,18], and the TS for the alkyl migratory insertion onto the coor-
dinated CO, have been considered. Although along the branched
profiles for the addition of the fourth CO group to the alkyl-Rh
tricarbonyl intermediates there are remarkably high barriers (see
Fig. 2 and Section 2.3.2), unlike for the linear profiles [19] discussed
below, the alkyl migratory insertion TS are higher in any case than
the CO addition ones. Therefore, in what follows, the CO addition
TS are disregarded when comparing the TS relative free energies,
but not when the reaction profiles are plotted.
Conformer names are 0, 1, 2 if the H2CCH3 value (Fig. 3) is
roughly trans (t), gauche+ (g), gauche− (gꢀ). Both enantiomers have
been considered for chiral substrates (four reactant intermediate
structures are reported in Fig. S1 of Electronic Supporting Informa-
tion (ESI)).
In one of the branched pathway profiles for TMP1 (Fig. 2), the
migratory insertion of the alkyl onto the coordinated CO leads
to a CO insertion TS higher than the alkyl-Rh TS (13.24 and
10.88 kcal/mol, respectively). Therefore the reaction proceeds to
the left, via ␤-hydride elimination, returning to the initial reactant
intermediate (Reac Int), which corresponds to the monodeuterated
terminal olefin TMP1-1-d1 found in the deuterioformylation exper-
iment at partial conversion. This occurs for one of the most stable
reactant complexes (named Reac0) taken as zero. The conformer
stability for intermediates should be a minor detail provided their
interconversion barriers are lower than the reaction TS according to
the Curtin–Hammet principle [20]. Conversely, in this case, neither
Reac2 (the reactant intermediate corresponding to the B2 alkyl-
Rh(CO)3 intermediate) nor the corresponding alkyl-Rh TS could
be located despite several attempts making use of constrained
Fig. 2. Free energy profile along one of the branched pathways for TMP1 (0-B2). The ␤-elimination TS leading to the internal olefin is drawn with dotted lines and empty
markers.

6
R. Lazzaroni et al. / Journal of Molecular Catalysis A: Chemical 356 (2012) 1–13
Fig. 3. Structures of the stationary points along the branched pathway for 0-B2 in Fig. 2. Their relative free energies (kcal/mol) and the TS imaginary frequencies (cm−1) are
also reported.
optimizations [19]. Interestingly, following the reaction pathway
backwards starting from the B2 alkyl-Rh(CO)3 intermediate, the B0
alkyl-Rh TS was obtained yielding Reac0, i.e. the most stable reac-
tant intermediate. This means also that when the B0 alkyl-Rh TS
is reached there are at least two competing branched pathways
leading to either the B0 or B2 alkyl-Rh(CO)3 intermediate. This is
the reason why in the legend of Fig. 2 the involved conformer has
been indicated as 0-B2. In this case the internal olefin cannot be
obtained, again via ␤-hydride elimination, because the TS for its
formation, also displayed in Fig. 2 (dotted line) for comparison, is
higher (13.66 kcal/mol) and thus less favorable than both alkyl-
Rh and CO insertion TS. Stationary-point structures through the
CO-insertion TS are displayed in Fig. 3.
The hydroformylation reaction for TMP1 is actually non-
reversible for Reac0 along the B0 pathway producing the
corresponding aldehyde, as occurs along the linear pathways for
all reactant complexes (one of the linear profiles is displayed in
Fig. 42: once the rate-limiting step, i.e. the alkyl-Rh TS, is crossed,
the reaction proceeds by formation of products).
Thus, from a computational view-point, the branched alkyl in
part produces again the starting olefin that reacts with the same
regioselectivity as before explaining the further increase in the lin-
ear fraction, and in part produces the branched aldehyde.
2
All the TS and intermediates along the profile are displayed for clarity, although
Interestingly enough, ␤-hydride elimination to yield the inter-
nal olefin occurs along the Bꢀ0 pathway for one of the reactant
complex considered (Reacꢀ0).
for Curtin–Hammett systems “theoretical calculation of the relevant transition-state
energies may avoid the task of considering ground-state isomer interconversion and
populations entirely” [20c].

R. Lazzaroni et al. / Journal of Molecular Catalysis A: Chemical 356 (2012) 1–13
7
behavior between branched and linear paths is striking. The
linear conformers cannot give the internal olefin, and the TS
for the migratory insertion of the alkyl onto the coordinate CO
has a free energy comparable with that of the reactant inter-
mediate (Fig. 5d). Concerning the branched conformers, only
B0 yields the aldehyde (Fig. 5c) as the linear ones, whereas
Bꢀ0 (Fig. 5b) ␤-eliminates producing the internal olefin, and
B2 (Fig. 5a and Fig. 2) ␤-eliminates producing the terminal
olefin.
2.3.2. Agostic interaction
The high barriers for the addition of the fourth CO group to the
alkyl-Rh tricarbonyl intermediates along the branched profiles are
due to the large stabilization of branched tricarbonyl intermediates
when an agostic interaction is established between Rh and one of
the nearby hydrogens. Examples of such interactions are shown in
Fig. 6 for the B2 alkyl-Rh(CO)3 intermediates deriving respectively
from TMP1 and DMP1. In the latter case, due to the H3CCHiso
torsion, three conformers, named B2g, B2t, and B2gꢀ are located.
B2gꢀ does not present however any agostic interaction because
Fig. 4. Free energy profile along the L0 linear pathway for one of the TMP1 reactant
intermediates. The CO addition TS cannot be located because, in the presence of CO,
linear tricarbonyl intermediates directly evolve to tetracarbonyl intermediates.
˚
Hiso is farther apart from Rh (3.16 A). The separations between Rh
and the H in ␤ are also displayed in each conformer, because they
are important for the ␤-hydride elimination reaction to give the
internal olefin.
2.3.1. Kinetic isotope effects
Besides from geometrical parameters, reported in Table 4 (typ-
ically the distance between the metal (M) and the hydrogen is
Despite the analogy found in the experimental runs between
hydro- and deuterioformylations, the kinetic isotope effects (KIE)
have been taken into account. In this case, KIE is given by:
˚
1.8–2.3 A and the M
H
C angle is 90–140◦), this kind of interaction
can be put forward from the existence of bond (BCP) and ring (RCP)
critical points in Bader’s topological analysis of the electron distri-
bution [16]. The critical points involving Rh and the delocalization
indeces for the atoms bonded to Rh are reported in Tables 5 and 6.
The graphical output obtained with AIMAll for tBu Bꢀ0 is displayed
in Fig. S2 of ESI. From a perusal of Tables 4–6 it is evident that the
agostic interactions (even those just outside the aforementioned
distance limit) are confirmed by QTAIM that finds BCP and RCP with
a significant density and delocalization index in all cases reported
but iso B2gꢀ and iso Bꢀ2gꢀ (Figs. S3b,d–S4 of ESI), where the iPr H
(Hiso) points toward Rh instead of one of those belonging to its
methyl groups (Fig. S5 of ESI).
It is noteworthy, however, that the behavior of this kind of sys-
tems is very similar at least concerning the agostic interaction.
Furthermore, as already remarked, the approach of the fourth CO
that might be hampered by a strong agostic interaction, shows
lower TS than the alkyl migratory insertion onto the coordinated
CO in any case.
kH
KIE =
kD
where kH and kD are the reaction rate constants, computed via the
Eyring–Polanyi equation [21]:
ꢀ
ꢁ
kBT
h
=/
/RT
k =
e−ꢁG
where k is the reaction rate constant, kB the Boltzmann constant, T
the absolute temperature, h the Planck’s constant, ꢀG =/ the Gibbs
free energy of activation, and R the gas constant.
To this aim, the free energies of all TS leading to the inter-
mediates in Scheme 2 have been computed with the proper
replacements of D to H: the differences are made with respect to
the lowest free energy reactant intermediates taken as zero.3
When the isotopic substitution is in the chemical bond that is
broken, i.e. in Rh-D as for the Rh-alkyl TS, the primary isotope effect
is large (1.69 ≤ KIE ≤ 1.71 (for L regioisomers)), as expected. Con-
versely, the secondary kinetic isotope effect (SKIE), observed when
D is contained in the alkyl part, not involved in bond breaking or
forming events, is much smaller ranging from 0.36 (for L regioiso-
mers) to 0.40. The velocity for the release of Rh-D is more than four
times slower than that for the release of Rh-H, as pointed out in
Section 3.
The results obtained for the monodeuterated TS are plotted in
Fig. 5 as compared to those of the species containing only H atoms.
From Fig. 5 it appears evident that the trend of hydroformylation
and deuterioformylation is very similar. Of course, in the case of
hydroformylation, the Rh-alkyl TS and the ␤-hydride elimination
TS to return to the terminal olefin (i.e. the initial reactant interme-
diate) coincide, and thus they are displayed twice, perfectly aligned,
in the plots. The TS for the release of D from the catalyst to the
substrate is higher than that for the release of H from the substrate
to the catalyst, as well known and noticed above. The different
2.3.3. Comparison between the hydroformylation behavior of
TMP1 (tBu) and DMP1 (iso)
The free energy based L:B regioselectivity ratios computed for
TMP1 and DMP1 under the hypothesis of nonreversibility (85:15
and 68:32, respectively) are significantly different, despite the
similarity between the two substrates. The trivial reason for the
computed ratios is that in the case of DMP1 there are 18 lin-
ear and 18 branched alkyl-Rh TS, whereas for TMP1 there are
6 linear and just 5 branched alkyl-Rh TS. The branched alkyl-
Rh TS corresponding to the tBu-B2 alkyl-Rh(CO)3 intermediate
was not reachable from the tBu-B2 alkyl-Rh(CO)3 intermediate,
because invariably the TS optimization produced the tBu-B0 alkyl-
Rh TS. Conversely, in an analogous run for DMP1, even the critical
B2gꢀ alkyl-Rh TS and the relevant reactant intermediate have
been obtained (Fig. 7). It is obvious that such structures are not
attainable for TMP1, ought to the steric hindrance of the tBu
group.
3
In order to clarify matters, the subsequent reaction steps have
been investigated also for DMP1 with a large computational effort
Reference free energies for Reac(H) and Reac(D) intermediates are −766.738577
and −766.741097 Eh, respectively. The free energy for CO is −113.560714 Eh.

8
R. Lazzaroni et al. / Journal of Molecular Catalysis A: Chemical 356 (2012) 1–13
Table 4
Geometrical parameters for some branched tricarbonyl intermediates of TMP1 (tBu) and DMP1 (iso).
iso B2gꢀ tBu Bꢀ0
3.8769 2.1618
tBu B2
2.4397
iso B2g
2.2909
iso B2t
2.3087
iso Bꢀ0t
2.1945
iso Bꢀ0g
iso Bꢀ2gꢀ
iso B0gꢀ
2.5710
Rh· · ·Ha (Å)
2.4391
123.92
3.5000
92.54
Rh· · ·H· · ·C (deg)
112.56
131.87
122.70
106.07 133.46
128.46
113.39
Fig. 5. TS free energies for a number of monodeuterated TMP1 species (solid markers) in comparison to those of the corresponding non-deuterated species (empty markers):
(a) B2D vs. B2; (b) Bꢀ0D vs. Bꢀ0; (c) B0D vs. B0; and (d) LD vs. L (all linear conformers show a very similar behavior); the lowest free energy reactant intermediates are taken
as zero.
Fig. 6. Agostic interactions in branched tricarbonyl intermediates obtained for TMP1 (B2) and DMP1 (B2g and B2t). The separation between Rh and the H at the ␤ carbon is
also displayed. Distances in Å.

R. Lazzaroni et al. / Journal of Molecular Catalysis A: Chemical 356 (2012) 1–13
9
Table 5
Electron density of bond (3, −1) and ring (3, +1) critical points for some branched tricarbonyl intermediates of TMP1 (tBu) and DMP1 (iso).
CP
tBu B2
iso B2g
iso B2t
iso B2gꢀ
tBu Bꢀ0
iso Bꢀ0t
iso Bꢀ0g
iso Bꢀ2gꢀ
iso B0gꢀ
C2-Rh
C4-Rh
C6-Rh
C8-Rh
Ha-Rh
C2-Rh-Ha-C21-C20-C14
0.1023
0.1252
0.1401
0.1408
0.0204
0.0124
0.0999
0.1277
0.1388
0.1401
0.0262
0.0108
0.1035
0.1243
0.1406
0.1410
0.0252
0.0126
0.1020
0.1277
0.1403
0.1399
–
0.0996
0.1279
0.1421
0.1357
0.0336
0.0122
0.1011
0.1266
0.1411
0.1382
0.0314
0.0126
0.0991
0.1291
0.1410
0.1368
0.0203
0.0170
0.1041
0.1271
0.1432
0.1374
–
0.0990
0.1294
0.1371
0.1403
0.0162
0.0154
–
–
Table 6
Delocalization index (DI(A,B)) of bonded atoms to Rh for some branched tricarbonyl intermediates of TMP1 (tBu) and DMP1 (iso).
CP
tBu B2
iso B2g
iso B2t
iso B2gꢀ
tBu Bꢀ0
iso Bꢀ0t
iso Bꢀ0g
iso Bꢀ2gꢀ
iso B0gꢀ
C2-Rh
C4-Rh
C6-Rh
C8-Rh
Ha-Rh
0.7876
1.0429
1.1058
1.1073
0.0878
0.7772
1.0636
1.0733
1.0941
0.1227
0.7946
1.0346
1.1048
1.1194
0.1071
0.7712
1.0699
1.0993
1.0979
–
0.7617
1.0695
1.1086
1.0542
0.1496
0.7696
1.0583
1.1081
1.1082
0.1374
0.7536
1.0824
1.0864
1.0575
0.0941
0.7767
1.0652
1.1364
1.0919
–
0.7528
1.0857
1.0593
1.0818
0.0721
Fig. 7. Reac2gꢀ intermediate (left) and B2gꢀ alkyl-Rh TS (right) obtained for DMP1. Distances in Å.
due to the number of steps and of conformers.4 The relative free
energy results for DMP1 are reported in Table 7, limited to the most
important TS5 relevant to reversibility. Analogous results for TMP1
are reported in Table 8 for comparison.
nonreversible and the branched aldehyde is produced opposite to
what occurs for 0-B2 TMP1.
The L0gꢀ free energy profile for DMP1, displayed in Fig. 9, closely
resembles that for L0 TMP1 shown in Fig. 4. Interested readers can
find the DMP1 structures in ESI (Figs. S6–S8).
In order to highlight the different behavior of 0-B2 and B2gꢀ
branched pathways for TMP1 and DMP1, respectively, the B2gꢀ
free energy profile for DMP1 is shown in Fig. 8. In this case, both
the Alkyl-Rh TS and the ␤-elim TS to yield the internal olefin are
higher than the CO-ins TS. Therefore the branched pathway is
From a perusal of Tables 7 and 8, it is indeed evident that
all linear conformers proceed to give the aldehyde products. An
analogous outcome is obtained for all the branched DMP1 con-
formers. The branched TMP1 conformers (Table 8) produce the
branched aldehyde with the exception of Bꢀ0 and B2 whose reac-
tion is reversible: Bꢀ0 undergoes ␤-hydride elimination to give the
internal olefin, whereas B2 undergoes ␤-hydride elimination to
return to the terminal olefin complex with the catalyst (see also
Figs. 2 and 5), i.e. to the starting material. This reacts again going
through the linear and branched paths with the original regiose-
lectivity, therefore increasing the linear fraction. This explains the
different behavior of DMP1 and TMP1.
4
As mentioned above, there are 18 linear and 18 branched structures for each
stationary point. A few constrained runs followed by a relaxed TS optimization are
necessary for each TS. To locate the lowest free energy CO insertion TS the three
equatorial CO groups need to be taken into account in turn also computing their
vibrational frequencies.
5
For CO insertion TS, only the lowest free energy ones are reported. Concerning
Solvent (benzene) effects had been considered for TMP1 [19]
exploiting either the continuum solvent in the IEF-PCM framework
[22] or the coordinating effect of a single solvent molecule. A
the reductive elimination to produce the aldehyde, only the B0 conformers have
been reported because the free energy of that species in TMP1 was higher than the
TSinsCO one and close enough to the alkyl-Rh TS one.

10
R. Lazzaroni et al. / Journal of Molecular Catalysis A: Chemical 356 (2012) 1–13
Table 7
Relative free energiesa (kcal/mol) for a number of TS along the branched and linear pathways for the hydroformylation of DMP1. Their imaginary frequencies (cm−1) are also
reported in parenthesis.
Conformer
Alkyl-Rh TS
CO-ins TS
Elim TSb
␤-elim TSc (internal)
B2g
B2gꢀ
B2t
B0g
B0gꢀ
B0t
B1g
B1gꢀ
B1t
18.76 (716 i)
13.02 (730 i)
15.79 (737 i)
10.27 (685 i)
9.92 (698 i)
10.77 (722 i)
14.20 (701 i)
14.04 (711 i)
12.02 (708 i)
11.44 (743 i)
11.40 (711 i)
12.22 (710 i)
9.79 (695 i)
13.62 (700 i)
14.08 (727 i)
15.65 (701 i)
19.08 (708 i)
15.82 (707 i)
10.68 (214i)
7.71 (220i)
9.90 (204i)
5.88 (219i)
3.22 (226i)
6.48 (228i)
9.33 (184i)
7.41 (218i)
5.89 (185i)
5.81 (244i)
6.78 (222i)
5.84 (201i)
3.17 (215i)
7.59 (199i)
12.63 (667i)
8.70 (652i)
13.68 (657i)
12.63 (667i)
8.70 (652i)
13.68 (657i)
6.97 (723i)
7.53 (711i)
4.80 (716i)
8.06 (715i)
Bꢀ2g
Bꢀ2gꢀ
Bꢀ2t
Bꢀ0g
Bꢀ0gꢀ
Bꢀ0t
Bꢀ1g
Bꢀ1gꢀ
Bꢀ1t
7.58 (649i)
8.96 (632i)
8.79 (645i)
7.58 (649i)
8.96 (632i)
8.79 (645i)
d
–
6.89 (192i)
10.99 (177i)
8.82 (196i)
L2g
L2gꢀ
L2t
L0g
L0gꢀ
L0t
L1g
L1gꢀ
L1t
16.07 (698i)
12.08 (688i)
14.88 (664i)
10.15 (694i)
9.49 (659i)
9.98 (682i)
11.65 (645i)
11.89 (660i)
10.80 (681i)
10.42 (678i)
10.24 (681i)
10.79 (666i)
9.69 (643i)
11.52 (651i)
11.97 (654i)
13.37 (620i)
16.01 (664i)
14.04 (657i)
5.67 (240i)
1.52 (252i)
2.86 (253i)
0.91 (273i)
0.91 (272i)
1.14 (263i)
4.20 (264i)
4.04 (270i)
5.38 (258i)
8.09 (241i)
8.22 (241i)
2.54 (272i)
1.54 (274i)
1.46 (262i)
1.46 (274i)
2.48 (274i)
3.06 (266i)
3.09 (264i)
3.42 (726i)
Lꢀ2g
Lꢀ2gꢀ
Lꢀ2t
Lꢀ0g
Lꢀ0gꢀ
Lꢀ0t
Lꢀ1g
Lꢀ1gꢀ
Lꢀ1t
a
Reference free energies are: Reac Int = −727.307455Eh, Reac Int + CO = −840.868169Eh, Reac Int + CO + H2 = −842.080705Eh.
b
c
Reductive elimination TS to produce aldehyde intermediates have been computed just for critical cases, when possibly competitive with relatively low alkyl-Rh TS.
While B1 and Bꢀ1 cannot ␤-eliminate as such because their H3 is on the opposite side with respect to Rh, B2 and B0 yield the same internal olefin, in analogy to Bꢀ2 and
Bꢀ0, because their H3 move (from gauche− and t) to a syn position with respect to Rh. Linear isomers cannot give the internal olefin via ␤-elimination, because Rh is on the
first C.
d
The relaxed TS optimizations invariably return to a tetracarbonyl intermediate structure with a rotated methyl group.
negligible effect on structure and free energies was observed in
the case of PCM, as expected, due to the small difference between
benzene (ε = 2.247) and vacuum dielectric constants. On the other
hand, in the supermolecule approach, the TS free energy gap
remained unaltered upon inclusion of a benzene molecule that did
not affect the mutual position of the various TS.
3. Experimental
Benzene was dried over molecular sieves and distilled under
nitrogen. Rh4(CO)12 was from Strem products. The starting olefins
(3-methylbut-1-ene and 3-methylpent-1-ene) were commercially
available. NMR spectra were recorded on a Varian Gemini 200 MHz
Table 8
Relative free energiesa (kcal/mol) for a number of TS along the branched and linear pathways for the hydroformylation of TMP1. Their imaginary frequencies (cm−1) are also
reported in parenthesis.
Conformer
Alkyl-Rh TS
CO-ins TS
Elim TS
␤-elim TSb (internal)
0-B2
B0
10.88 (689i)
10.88 (689i)
15.02 (702i)
11.69 (720i)
14.52 (697i)
18.53 (721i)
13.24 (212i)
8.16 (209i)
9.47 (179i)
7.41 (227i)
9.91 (190i)
11.70 (173i)
10.56 (737i)
9.66 (710i)
7.60 (735i)
6.83 (705i)
8.32 (706i)
14.02 (719i)
13.66 (680i)
13.72 (685i)
–
9.48 (630i)
9.48 (630i)
–
B1
Bꢀ2
Bꢀ0
Bꢀ1
L2
L0
L1
17.61 (684i)
9.80 (682i)
12.41 (645i)
10.90 (679i)
13.03 (651i)
17.98 (554i)
2.73 (277i)
1.36 (265i)
4.41 (265i)
1.20 (277i)
2.14 (262i)
4.48 (273i)
6.16 (714i)
3.31 (718i)
6.02 (716i)
2.76 (720i)
4.21 (719i)
5.72 (725i)
–
–
–
–
–
–
Lꢀ2
Lꢀ0
Lꢀ1
a
Reference free energies are: Reac Int = −766.738577Eh, Reac Int + CO = −880.299291Eh, Reac Int + CO + H2 = −881.511827Eh.
b
While B1 and Bꢀ1 cannot ␤-eliminate as such because their H3 is on the opposite side with respect to Rh, B2 and B0 yield the same internal olefin, in analogy to Bꢀ2 and
Bꢀ0, because their H3 move (from gauche− and t) to a syn position with respect to Rh. Linear isomers cannot give the internal olefin via ␤-elimination, because Rh is on the
first C.

R. Lazzaroni et al. / Journal of Molecular Catalysis A: Chemical 356 (2012) 1–13
11
Fig. 8. Free energy profile along the B2gꢀ branched pathway for DMP1 corresponding to the 0-B2 for TMP1. The ␤-elimination TS leading to the internal olefin is drawn with
dotted lines and empty markers.
Fig. 9. Free energy profile along the L0gꢀ linear pathway for one of the DMP1 reactant intermediates (Reac0gꢀ).
and on a Varian Gemini 300 MHz using benzene-d6 as solvent. GC
analysis was carried out on a Perkin Elmer 8600 gas chromatograph,
using a BP1 column (15 m × 0.32 mm × 0.25 ␮m), using helium as
carrier gas.
and hydrogen or deuterium was rapidly introduced up to 100 atm
total pressure (CO/H2(D2) = 1:1). When the gas absorption reached
the value corresponding to the fixed conversion, the reaction mix-
ture was siphoned out. The degree of conversion was measured by
GLC, using n-decane as the internal standard.
3.1. Hydroformylation or deuterioformylation of
3,4,4-trimethylpent-1-ene (TMP1): general procedure
4. Computational details
A
solution of 3,4,4-trimethylpent-1-ene (2.78 mmol) and
The Rh-carbonyl hydride [H Rh(CO)3], produced by the unmod-
ified catalytic precursor [Rh4(CO)12] under mild hydroformylation
conditions, is the catalytic active species used to study the reaction
regioselectivity (L:B, Scheme 1). Since the reaction occurs at rt, a
Rh4(CO)12 (Rh/substrate = 1/1000) in benzene (5 ml) was intro-
duced by suction into an evacuated 25 ml stainless steel autoclave.
Carbon monoxide was introduced, the autoclave was then rocked

12
R. Lazzaroni et al. / Journal of Molecular Catalysis A: Chemical 356 (2012) 1–13
consistent comparison between theoretical and experimental val-
ues can be performed, as we demonstrated in our first study on the
hydroformylation regioselectivity [13a] for a series of eight mono-
substituted olefins plus ethene and dimethylethene. Because for
nonreversible reactions the olefin insertion into the Rh H bond
turned out to be the step determining the regioselectivity,6 in
the computational strategy employed, the L:B regioisomeric ratio
was theoretically evaluated exploiting the free energy differences
that can be approximated by the potential energy differences, of
branched and linear alkyl-rhodium transition states, using the for-
mula:
group was carried out. In all cases the linear isomer prevails with
respect to the branched one in the 2:1 ratio or more. An anoma-
lous behavior is found for TMP1 (R = tBu), where the linear isomer
largely prevails (L:B = 95:5). A profound examination of the reaction
course via deuterioformylation experiments (carried out at partial
substrate conversion as well) showed that the alkyl-rhodium for-
mation resulted to be irreversible for all substrates except for the
bulkier one (TMP1). In that particular case, a significant amount of
␤-elimination was observed but only for the branched alkyl isomer.
This fact explains the high selectivity in favor of the linear aldehyde
because, while the linear alkyl proceeds toward the aldehyde, the
branched alkyl primarily undergoes ␤-elimination and only in a
very limited amount is transformed into the related aldehyde. It
should be noted that the 2H NMR analysis of the crude deuteri-
oformylation reaction mixture is a direct and simple method to
investigate the different nature and fate of the alkyl–metal inter-
mediates [3,7,8,31–34].
Theoretical calculations on the regioselectivity of the
alkyl–metal intermediates formation show a good agreement
with regioselectivity values experimentally determined on the
aldehyde. Only in the case of TMP1, the theoretical values of
regioselectivity (L:B = 85:15), calculated under the hypothesis
of nonreversibility, are slightly underestimated with respect to
the experimental values (L:B = 95:5). Interestingly, the careful
analysis of the branched pathway put forward a free energy
transition state (TS) for the migratory insertion of the alkyl onto
the Rh-coordinated CO higher than the alkyl-Rh formation TS thus
preventing the branched aldehyde formation. In such a case, the
reaction moves backwards, producing again the initial olefin com-
plex (corresponding to the monodeuterated TMP1-1-d1 species
obtained in deuterioformylation experiments at partial substrate
conversion) and enriching the linear fraction that invariably
proceeds to aldehyde. An analogous analysis carried out for DMP1
puts forward a nonreversible mechanism for every conformer and
points out that the steric hindrance of the tBu group hampers
the attainment of the B2 alkyl-Rh TS for TMP1 favoring the initial
olefin complex formation for that conformer.
ꢂ
ꢂ
ꢂ
=/
=/
=/
/RT
/RT
/RT
L
B
L
L : B = kL : kB
=
e−ꢀG
:
e−ꢀG
≈
e−ꢁE
ꢂ
=/
/RT
:
e−ꢁE
(1)
B
where k is the reaction rate, ꢀG =/ and ꢀE =/ are the relative TS
free energies and potential energies with respect to the most stable
TS. Zero-point vibrational energies and thermal corrections were
computed in the rigid rotor-harmonic oscillator approximation to
obtain the free energies [23] at 298.15 K. The summations run over
all the possible branched and linear alkyl-Rh TS structures.
Density functional theory (DFT) calculations have been carried
out with the Gaussian 03 suite of codes [24], using for C, O, and H the
Becke gradient-corrected three-parameter hybrid exchange and
Perdew 86 gradient-corrected correlation functionals, i.e. B3P86
[25,26], and the 6-31G* basis set [27]. Effective core potentials
(implicitly including some relativistic effects for the electrons near
the nucleus), coupled to the LanL2DZ valence basis set, have been
used for Rh [28].
Due to the conformational flexibility of 1-alkenes, a variety of
reactant complexes with H-Rh(CO)3 must be considered to explore
the reaction mechanism. For alkyl-Rh TS all rotamers need to be
taken into account although only those within ∼5 kcal/mol of the
most favorable one are included into Eq. (1) to evaluate the regios-
electivity [13]. Branched and linear pathways must be followed
for each conformer when subsequent reaction steps are exam-
ined, because it is difficult to model build from scratch likely TS
structures. Several constrained optimizations in the TS region are
necessary that are eventually relaxed employing TS optimizations.
In few cases however it was impossible to obtain the sought TS
structures despite a number of trials, because TS optimizations lead
to rotational TS of nearby intermediates.
Therefore, although the steric hindrance was expected to influ-
ence the reaction, it was however difficult to understand how
and where from the experiment. While deuterioformylation runs
are useful to show that ␤-hydride elimination occurs, only the
theoretical inspection explains at which step of the reaction this
happens.
QTAIM calculations have been carried out with AIMAll [29].
Structures and normal modes have been visualized using Molden
[30].
This investigation clearly illustrates how important (perhaps
fundamental) is to tackle regioselectivity issues in hydroformy-
lation reactions using different and complementary approaches,
such as deuterioformylations and theoretical calculations, because
together they permit a better understanding and a proper inter-
pretation of the mechanism of this significant reaction which
despite over 60 years old becomes more and more beautiful and
intriguing.
5. Conclusions
It is well documented [7] that the olefin structure plays an
important role in the reaction regioselectivity. The branched iso-
mer is remarkably more favored when vinyl aromatic substrates,
such as vinylpyrrole, vinylstyrene, or vinylpyridine, are taken into
account. The branched isomer also prevails in the hydroformyla-
tion of vinyl allyl ethers. In the case of the alkyl vinyl and alkyl allyl
substrates both regioisomers are formed in a very similar amount.
In order to evaluate how the steric hindrance due to a sec-alkyl
group bonded to the vinyl group can affect the ␤-regioselectivity of
the reaction, the hydroformylation of four substrates (3-methylbut-
1-ene (MB1), 3-methylpent-1-ene (MP1), 3,4-dimethylpent-1-ene
(DMP1), and 3,4,4-trimethylpent-1-ene (TMP1)), characterized by
an increasing steric hindrance of the group bonded to the vinyl
Acknowledgments
R.L. and R.S. are indebted to Prof. Aldo Vitagliano for supplying
the 3,4-dimethylpent-1-ene and 3,4,4-trimethylpent-1-ene sam-
ples. G.A. and C.G. are grateful to Todd A. Keith for practical advice
and helpful comments on the treatment of ECP-modeled core elec-
trons in AIMAll.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.molcata.2011.12.021.
6
That step can determine also the diastereoselectivity as in vinyl olefins contain-
ing chiral alkoxy or alkyl groups [13b].

R. Lazzaroni et al. / Journal of Molecular Catalysis A: Chemical 356 (2012) 1–13
13
References
[21] (a) H. Eyring, J. Chem. Phys. 3 (1935) 107–115;
(b) M.G. Evans, M. Polanyi, Trans. Faraday Soc. 31 (1935) 875–894.
[22] (a) S. Miertus, E. Scrocco, J. Tomasi, Chem. Phys. 55 (1981) 117–129;
(b) S. Miertus, J. Tomasi, Chem. Phys. 65 (1982) 239–245;
[1] R. Lazzaroni, R. Settambolo, G. Uccello-Barretta, Organometallics 14 (1995)
4644–4650.
[2] (a) B.E. Hanson, N.E. Davis, J. Chem. Educ. 64 (1987) 928–931;
(b) R. Lazzaroni, P. Pertici, S. Bertozzi, G. Fabrizi, J. Mol. Catal. 58 (1990) 75–85.
[3] R. Lazzaroni, G. Uccello-Barretta, M. Benetti, Organometallics
2323–2327.
[4] I. Ojima, Chem. Rev. 88 (1988) 1011–1030.
[5] (a) P. Kalck, F. Serein-Spirau, New J. Chem. 13 (1989) 515–518;
(c) J. Tomasi, M. Persico, Chem. Rev. 94 (1994) 2027–2094;
(d) J. Tomasi, B. Mennucci, E. Cancès, J. Mol. Struct. Theochem. 464 (1999)
211–226;
8 (1989)
(e) J. Tomasi, B. Mennucci, R. Cammi, Chem. Rev. 105 (2005) 2999–3094.
[23] D.A. McQuarrie, Statistical Mechanics, University Science Book, Sausalito, CA,
2000.
[24] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheese-
man, J.A. Montgomery Jr., T. Vreven, K.N. Kudin, J.C. Burant, J.M. Millam, S.S.
Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G.A.
Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa,
M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J.E. Knox, H.P.
Hratchian, J.B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R.E. Strat-
mann, O. Yazyev, A.J. Austin, R. Cammi, C. Pomelli, J.W. Ochterski, P.Y. Ayala, K.
Morokuma, G.A. Voth, P. Salvador, J.J. Dannenberg, V.G. Zakrzewski, S. Dapprich,
A.D. Daniels, M.C. Strain, O. Farkas, D.K. Malick, A.D. Rabuck, K. Raghavachari,
J.B. Foresman, J.V. Ortiz, Q. Cui, A.G. Baboul, S. Clifford, J. Cioslowski, B.B. Ste-
fanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R.L. Martin, D.J. Fox, T. Keith,
M.A. Al-Laham, C.Y. Peng, A. Nanayakkara, M. Challacombe, P.M.W. Gill, B. John-
son, W. Chen, M.W. Wong, C. Gonzalez, J.A. Pople, Gaussian 03, Revision C. 02,
Gaussian Inc, Wallingford, CT, 2004.
(b) A.L. Lapidus, A.P. Rodin, I.G. Pruidze, B.I. Ugrak, Izv. Akad. Nauk SSSR, Ser.
Khim 7 (1990) 1661–1662.
[6] A.F. Browning, A.D. Bacon, C. White, D.J. Milner, J. Mol. Catal. 83 (1993) L11–L14.
[7] R. Lazzaroni, R. Settambolo, A. Caiazzo, in: P.W.N.M. van Leeuwen, C. Claver
(Eds.), Rhodium Catalyzed Hydroformylation, Kluwer, Dordrecht, 2000, Chap-
ter 2.
[8] R. Settambolo, S. Scamuzzi, A. Caiazzo, R. Lazzaroni, Organometallics 17 (1998)
2127–2130.
[9] A. Caiazzo, R. Settambolo, L. Pontorno, R. Lazzaroni, J. Organomet. Chem. 599
(2000) 298–303.
[10] C.P. Casey, L.M. Petrovich, J. Am. Chem. Soc. 117 (1995) 6007–6014.
[11] N. Sakai, S. Mano, K. Nozaki, H. Takaya, J. Am. Chem. Soc. 115 (1993) 7033–7034.
[12] K. Nozaki, N. Sakai, T. Nanno, T. Higashijima, S. Mano, T. Horiuchi, H. Takaya, J.
Am. Chem. Soc. 119 (1997) 4413–4423.
[13] (a) G. Alagona, C. Ghio, R. Lazzaroni, R. Settambolo, Organometallics 20 (2001)
5394–5404;
[25] A.D. Becke, J. Chem. Phys. 98 (1993) 5648–5652.
[26] J.P. Perdew, Phys. Rev. B 33 (1986) 8822–8824.
[27] W.J. Hehre, L.P. Radom, P.v.R. Schleyer, J.A. Pople, Ab Initio Molecular Orbital
Theory, Wiley, New York, 1986.
[28] P.J. Hay, W.R. Wadt, J. Chem. Phys. 82 (1985) 270–283.
[29] Molden (Version 4.8), G. Schaftenaar, J.H. Noordik, J. Comput. Aided Mol. Des.
14 (2000) 123–134.
[30] T.A. Keith, AIMAll (Version 11.10.16), TK Gristmill Software, Overland Park, KS,
USA, 2011 (aim.tkgristmill.com).
[31] G. Uccello-Barretta, R. Lazzaroni, R. Settambolo, P. Salvadori, J. Organomet.
Chem. 417 (1991) 111–119.
[32] R. Lazzaroni, R. Settambolo, A. Caiazzo, M. Bennett, Organometallics 21 (2002)
2454–2459.
[33] R. Lazzaroni, R. Settambolo, G. Prota, C. Botteghi, S. Paganelli, M. Marchetti,
Inorg. Chim. Acta 357 (2004) 3079–3083.
[34] R. Lazzaroni, R. Settambolo, M. Marchetti, S. Paganelli, G. Alagona, C. Ghio, Inorg.
Chim. Acta 362 (2009) 1641–1644.
[35] D. Evans, J.A. Osborn, G. Wilkinson, J. Chem. Soc. A (1968) 3133–3142.
(b) G. Alagona, C. Ghio, R. Lazzaroni, R. Settambolo, Inorg. Chim. Acta 357 (2004)
2980–2988.
[14] G. Alagona, R. Lazzaroni, C. Ghio, J. Mol. Model. 17 (2011) 2275–2284.
[15] (a) H.C. Urey, D. Rittenberg, J. Chem. Phys. 1 (1933) 137–143;
(b) J. Bigeleisen, M.G. Mayer, J. Chem. Phys. 15 (1947) 261–267;
(c) J. Evans, J. Schwartz, P.W. Urquhart, J. Organomet. Chem. 81 (1974) C37.
[16] R.F.W. Bader, Atoms in Molecules: A Quantum Theory, Oxford University Press,
Oxford, 1990.
[17] C. Ghio, R. Lazzaroni, G. Alagona, Eur. J. Inorg. Chem. (2009) 98–103.
[18] R. Lazzaroni, R. Settambolo, G. Alagona, C. Ghio, Coord. Chem. Rev. 254 (2010)
696–706.
[19] G. Alagona, C. Ghio, Theor. Chem. Acc., in press.
[20] (a) D.Y. Curtin, Rec. Chem. Prog. 15 (1954) 111–128;
(b) L.P. Hammett, Physical Organic Chemistry, 2nd edn., McGraw-Hill, New
York, 1970, pp. 119–120, Chapter 5;
(c) J.I. Seeman, Chem. Rev. 83 (1983) 84–134.