Online monitoring of hydroformylation intermediates by ESI-MS

Article abstract of DOI:10.1021/om100131t

Self-assembling ligands bearing permanently charged moieties have been synthesized and investigated in the Rh-catalyzed hydroformylation of terminal alkenes. By coupling a high-pressure autoclave directly to an ESI mass spectrometer hydroformylation reactions applying self-assembling 6-DPPon ligands could be studied in an online fashion. The live-streaming of the reaction mixture to the spectrometer revealed a series of different complexes not observed by other methods before, the structures of which were corroborated by CID experiments. Under CO/H₂ atmosphere, new complexes that are predicted by the Wilkinson catalytic cycle could be identified and studied by CID experiments, too. Especially the ion at m/z 848, a square-planar hydrido-carbonyl complex that is normally not detectable by other methods, was investigated in detail. Collision experiments of this complex resulted in the loss of CO and H₂, the latter being quite unusual, and points to the involvement of the hydrogen bond framework. These findings were further supported by deuteration experiments that revealed a clear incorporation of deuterium into the ligands. From these findings a new hydrogen-activation mechanism was proposed. Furthermore, substrate-containing complexes could be generated too, though a huge excess of substrate was necessary. CID experiments either with D₂ or Ar yielded nearly identical spectra, hinting at a complex that might result either from a β-hydride elimination or from intramolecular oxidative addition of one of the ligands.

Full text of DOI:10.1021/om100131t

Organometallics 2010, 29, 2521–2532 2521
DOI: 10.1021/om100131t
Online Monitoring of Hydroformylation Intermediates by ESI-MS
Christian H. Beierlein and Bernhard Breit*
Institut fu€r Organische Chemie und Biochemie and Freiburg Institute for Advanced Studies (FRIAS),
Albert-Ludwigs-Universita€t Freiburg, Albertstrasse 21, 79104 Freiburg, Germany
Roberto A. Paz Schmidt and Dietmar A. Plattner*
€
Institut fu€r Organische Chemie und Biochemie, Albert-Ludwigs-Universitat Freiburg, Albertstrasse 21,
79104 Freiburg, Germany
Received February 19, 2010
Self-assembling ligands bearing permanently charged moieties have been synthesized and inves-
tigated in the Rh-catalyzed hydroformylation of terminal alkenes. By coupling a high-pressure
autoclave directly to an ESI mass spectrometer hydroformylation reactions applying self-assembling
6-DPPon ligands could be studied in an online fashion. The live-streaming of the reaction mixture to
the spectrometer revealed a series of different complexes not observed by other methods before, the
structures of which were corroborated by CID experiments. Under CO/H2 atmosphere, new
complexes that are predicted by the Wilkinson catalytic cycle could be identified and studied by
CID experiments, too. Especially the ion at m/z 848, a square-planar hydrido-carbonyl complex that
is normally not detectable by other methods, was investigated in detail. Collision experiments of
this complex resulted in the loss of CO and H2, the latter being quite unusual, and points to the
involvement of the hydrogen bond framework. These findings were further supported by deuteration
experiments that revealed a clear incorporation of deuterium into the ligands. From these findings
a new hydrogen-activation mechanism was proposed. Furthermore, substrate-containing com-
plexes could be generated too, though a huge excess of substrate was necessary. CID experiments
either with D2 or Ar yielded nearly identical spectra, hinting at a complex that might result either from
a β-hydride elimination or from intramolecular oxidative addition of one of the ligands.
Scheme 1. DNA Base Pair Inspired Self-Assembling Ligands
Introduction
In times of dwindling resources, transition-metal-cata-
lyzed transformations are more and more a focus of common
interest, enabling otherwise time- and energy-consuming
processes in much milder and cost-efficient ways. A thor-
ough understanding of the course and mechanism of these
chemical reactions is of profound importance for improving
catalysts and catalytic transformations by identifying reac-
tivity- and selectivity-controlling interactions, intermediates,
and transition states. However, due to the often unstable
and short-lived nature of those species of interest, detection
can be a tedious and time-consuming endeavor, applying,
for example, isotope-labeling experiments. Thus, online
methods that enable the direct observation of an ongoing
reaction like IR and NMR techniques are commonly used
nowadays. Unfortunately, specialized equipment is needed
in most cases and the setup tends to be complicated and
limited to only a handful of reactions, in addition to high
ownership costs (e.g., ReactIR). The required accuracy for
detection of intermediate species in low concentrations
is often not met, thus rendering the methodology useless
for online experiments. Because of the ease of transferring
solution-phase ions to the gas phase for identification or
further reactions, electrospray ionization mass spectrometry
(ESI-MS) has become increasingly popular as a mechanistic
tool for studying short-lived reactive intermediates in orga-
nometallic catalytic cycles.1,2 Extending the previous appli-
cations to investigate chemical processes under pressure by
ESI-MS, a setup for the direct coupling of a high-pressure
autoclave to an electrospray tandem mass spectrometer was
(1) Plattner, D. A. J. Mass Spectrom. 2001, 207, 125, and references
therein.
*Corresponding authors. E-mail: bernhard.breit@chemie.uni-freiburg.de;
dietmar.plattner@chemie.uni-freiburg.de.
(2) Plattner, D. A. Top. Curr. Chem. 2003, 225, 153, and references
therein.
r
2010 American Chemical Society
Published on Web 05/05/2010
pubs.acs.org/Organometallics

2522 Organometallics, Vol. 29, No. 11, 2010
Beierlein et al.
Scheme 2. Synthesis of 2
Scheme 3. ESI-MS Spectrum of 10 in CH3CN
described recently.3 This intriguing feature prompted us to
have a closer look into the Rh-catalyzed hydroformylation
of terminal alkenes,4 one of the most important industrial
applied transition-metal-catalyzed reactions and, by the
definition of Trost,5 a truly atom-economic transformation.
Recently, Krische and co-workers demonstrated the unique
advantages of ESI-MS for mechanistic studies on the related
coupling of acetylene to carbonyl compounds.6
As the preparation of bidentate ligands often incorpo-
rates elaborate multistep syntheses, one of us introduced
the concept of self-assembling ligands that mimic a biden-
tate binding motif in the presence of a transition metal
(Scheme 1).7,8 By simple mixing two monodentate ligands
bearing a complementary hydrogen bond donor-acceptor
motif, bidentate ligands can be formed in a combinatorial
fashion. Attractive hydrogen-bond interactions of the
donor-acceptor or acceptor-donor counterparts further
stabilize the resulting complexes, which were successfully
(3) Paz Schmidt, R. A.; Plattner, D. A. Anal. Chem. 2009, 81, 3665.

Article
Organometallics, Vol. 29, No. 11, 2010 2523
Scheme 4. 31P NMR Spectrum of 11
Scheme 5. 1H NMR spectrum of 11
Table 1. Hydroformylation of 1-Octene Applying 10, 6-DPPon,
and Mixtures of Ligandsa
applied in a wide range of chemical transformations, e.g.,
rhodium-catalyzed regioselective hydroformylation of terminal
(4) For selected reviews on alkene hydroformylation, see: (a) Cornils,
B.; Herrmann, W. A.; Kohlpaintner, C. W. Otto Roelen as the Pioneer of
Industrial Homogeneous Catalysis. Angew. Chem., Int. Ed. Engl. 1994,
33, 2144. (b) Beller, M.; Cornils, B.; Frohning, C. D.; Kohlpaintner, C. W.
Progress in Hydroformylation and Carbonylation. J. Mol. Catal. 1995, 104,
17. (c) Frohning, C. D.; Kohlpaintner, C. W. In Applied Homogeneous
Catalysis with Organometallic Compounds, Vol. 1; Cornils, B.; Herrmann,
W. A., Eds.; Wiley-VCH: Weinheim, 1996; pp 29-104. (d) Nozaki, K. In
Comprehensive Asymmetric Catalysis; Jacobsen, E. N., Pfaltz, A., Yamamo-
to, H., Eds.; Springer-Verlag: Berlin, 1999; Vol. 1, pp 381-413. (e) Breit, B.
Synthetic Aspects of Stereoselective Hydroformylation. Acc. Chem. Res.
2003, 36, 264. (f) van Leeuwen, P. W. N. M. Homogeneous Catalysis:
Understanding the Art; Kluwer: Dordrecht, 2004. (g) Breit, B. Recent
Advances in Alkene Hydroformylation. Top. Curr. Chem. 2007, 279, 139.
(5) Trost, B. M. Science 1991, 254, 1471.
entry
solvent
ligand
6-DPPon
10
6-DPPon/10 (1:1)
6-DPPon
conv (%)
selectivity (l:b)
1
CH3CN
CH3CN
CH3CN
toluene
100
100
100
100
93:7
94:6
95:5
97:3
2
3b
4
a Conditions: Rh:L:1-octene = 1:10:500, c(1-octene) = 0.36 M, CO/H2
(1:1), 10 bar, 70 °C, 20 h. b Rh:L1:L2:1-octene = 1:5:5:500, c(1-octene) =
0.36 M, CO/H2 (1:1), 10 bar, 70 °C, 20 h.

2524 Organometallics, Vol. 29, No. 11, 2010
Beierlein et al.
Scheme 6. ESI-MS Spectrum of 10 and [Rh(NBD)2]BF4 in CH3CN (m/z 200-1200)
alkenes,9 ruthenium-catalyzed anti-Markovnikov hydration of
terminal alkynes,10 rhodium-catalyzed asymmetric hydrogena-
tion,11 and nickel-catalyzed hydrocyanation of styrenes.12
The depicted 6-DPPon ligand 1 can be seen as the seminal
example of our self-assembling ligands and holds an exceptional
position, as it can act either as a hydrogen-donor-acceptor or
hydrogen-acceptor-donor ligand by fast tautomerization be-
tween the pyridone and the hydroxypyridine form.9 Though this
self-complementarity impairs its applicability in the combina-
torial generation of bidentate ligands, one has to synthesize only
one monodentate ligand to generate bidentate ligands. Bearing
this in mind, we have chosen the 6-DPPon platform as a model
system for this mechanistic study.
Scheme 7. Catalytic Cycle for the Rhodium-Catalyzed Hydro-
formylation of 1-Hexene with the Masses of Expected Complexes
Experimental Part
Ligand Synthesis. As only charged species can be detected by
means of ESI-MS, we had to ensure that at any time in the
catalytic cycle the intermediates of interest bear either a positive
(6) Williams, V. M.; Rock Kong, J.; Joon, Ko, B.; Mantri, Y.;
Brodbelt, J. S.; Baik, M.-H.; Krische, M. J. J. Am. Chem. Soc. 2009,
131, 16054.
(7) Breit, B. Angew. Chem. 2005, 117, 6976. Angew. Chem., Int. Ed.
2005, 44, 6816. Breit, B.; Seiche, W. J. Am. Chem. Soc. 2003, 125, 6608.
(8) For alternative approaches to bidentate ligands through self-
assembly, see ref 6 and: (a) Takacs, J. M.; Reddy, D. S.; Moteki, A.;
Wu, D.; Palencia, H. J. Am. Chem. Soc. 2004, 126, 4494. (b) Wilkinson,
M. J.; van Leeuwen, P. W. N. M.; Reek, J. N. H. Org. Biomol. Chem. 2005, 3,
€
2371. (c) Slagt, V. F.; Roder, M.; Kamer, P. C. J.; van Leeuwen, P. W. N. M.;
Reek, J. N. H. J. Am. Chem. Soc. 2004, 126, 4056. (d) Slagt, V. F.; van
Leeuwen, P. W. N. M.; Reek, J. N. H. Angew. Chem. 2003, 115, 5777. ;
Angew. Chem., Int. Ed. 2003, 42, 5619. (e) Slagt, V. F.; van Leeuwen,
P. W. N. M.; Reek, J. N. H. Chem. Commun. 2003, 2474.
(9) (a) Breit, B.; Seiche, W. J. Am. Chem. Soc. 2003, 125, 6608. (b) Breit,
B.;Seiche, W. Angew. Chem. 2005, 117, 1666. ; Angew. Chem., Int. Ed. 2005,
44, 1640. (c) Seiche, W.; Schuschkowski, A.; Breit, B. Adv. Synth. Catal.
2005, 347, 1488. (d) Waloch, C.; Wieland, J.; Keller, M.; Breit, B. Angew.
Chem. 2007, 119, 3097. ; Angew. Chem., Int. Ed. 2007, 46, 3037.
(10) Chevallier, F.; Breit, B. Angew. Chem. 2006, 118, 1629. ; Angew.
Chem., Int. Ed. 2006, 45, 1599.
or a negative charge. Hence, we prepared the 6-DPPon deriva-
tive 2, bearing an additional hydroxymethylene- moiety for
further functionalization, e.g., through esterification with a
charged carboxylic acid. This substitution pattern was chosen
to ensure minimal interaction between the additional functional
group and the hydrogen-bond network. Following literature
precedence,13 we decided to introduce a quaternary amine as a
functional group with a permanent positive charge, which
(11) Weis, M.; Waloch, C.; Seiche, W.; Breit, B. J. Am. Chem. Soc.
2006, 128, 4188.
(12) de Greef, M.; Breit, B. Angew. Chem. 2009, 121, 559.
(13) (a) Mohr, B.; Lynn, D. M.; Grubbs, R. H. Organometallics 1996,
15, 4317. (b) Hinderling, C.; Adhlart, C.; Chen, P. Angew. Chem., Int. Ed.
1998, 37, 2685.

Article
Organometallics, Vol. 29, No. 11, 2010 2525
Scheme 8. ESI-MS Spectrum of the 10/1/[Rh(CO)2acac] Mixture under a CO/H2 Atmosphere
Scheme 9. ESI-MS Spectrum of the Proposed Resting State A0
should exhibit a much higher ionization tendency than nega-
tively charged groups. The complete synthesis toward the final
ligand 10 is depicted in Scheme 2, and detailed experimental data
and spectra are given in the Supporting Information.
ESI-MS investigation of a CH3CN solution of 10 confirmed
the purity and the high ionization tendency of the charged
ligand, as anticipated (Scheme 3).
complex, with coupling constants greater than 3000 Hz suggest-
ing a cis arrangement of the ligands. Furthermore, the formation
of stabilizing hydrogen bonds can be monitored in the proton
NMR, normally as a pair of signals at around δ 10-12 ppm.
Thus, PtCl2(COD) was mixed with 2 equiv of 10 in CD3CN
and investigated by 31P/1H NMR spectroscopy. The resulting
spectra, depicted in Schemes 4 and 5, clearly show the formation
of a [PtCl2(10)2] complex (the obtained P,Pt coupling constant
of JP,Pt = 3645.78 Hz indicates a cis configuration), which cor-
responds to structure 11 and is in good agreement with the
results obtained for the parent system [PtCl2(6-DPPon)2].9
Comparison of the Performance of 10 and 6-DPPon (1) in the
Course of the Rhodium-Catalyzed Hydroformylation. To test the
ability of ligand 10 to form bidentate metal ligand complexes, a
[PtCl2(ligand)2] complex was generated in situ and studied by 31
P
and 1H NMR spectroscopy. The resulting P,Pt coupling constant
can provide an indication about the structure of the formed
In the H NMR, the two hydrogens involved in hydrogen
bonding show a resonance at δ = 12.1 ppm. The two distinct
1

2526 Organometallics, Vol. 29, No. 11, 2010
Scheme 10. CID Experiment of m/z 1127
Beierlein et al.
signals normally observed coincide to a broad signal with the
intensity of two protons, probably due to solvent effects.
To make sure that the ammonium-functionalized new mono-
dentate ligand 10 and the parent 6-DPPon system 1 display
similar self-assembling properties in catalysis, we studied both
systems in the course of the Rh-catalyzed hydroformylation em-
ploying 1-octene as a prototype for a terminal alkene (Table 1).
Due to solubility reasons, the reactions were carried out in
CH3CN as a solvent, which surprisingly did not influence the
catalyst performance with regard to selectivity or reactivity (see
entry 4 for a reaction in toluene that has been employed as the
standard solvent previously9). It is known that polar solvents
degrade the performance of these catalysts with regard to both
activity and selectivity.14
Scheme 11. X-ray Structure of 15
In all cases, conversion was quantitative after 20 h. The
obtained selectivities for the linear product were excellent for
both the cationic system 10 and 6-DPPon (1) and clearly indicate
the formation of a pseudo-bidentate catalyst system in all cases.
Furthermore, also when applying a 1:1 mixture of charged
and uncharged ligand, the observed high regioselectivity indi-
cates the formation of a self-assembling ligand system (Table 1,
entry 3). In conclusion, no difference could be observed between
6-DPPon (1) and the new cationic derivative 10.
ESI-MS Investigations. To shed some light on the mechanism
of the hydroformylation reaction, several steps were examined.
First, the catalyst precursor complex resulting from mixing the
ligand and [Rh(COD)2]BF4 was electrosprayed to assess the
complexity of the ion mixture; afterward online monitoring of
the reaction medium under hydroformylation conditions (both
with and without substrate) was performed.
Thus, a CH3CN solution of the catalyst precursor generated
in situ upon mixing 4 equiv of cationic ligand 10 and 1 equiv of
[Rh(NBD)2]BF4 was investigated by ESI-MS, revealing a much
more complex mixture of ions than expected (Scheme 6 and
Supporting Information).
Most of these ions have not been detected by other methods
before. IR and NMR experiments normally only show a single
complex that is thought to be a resting state containing three
ligands. ESI-MS experiments show, among other ions, a com-
plex consisting of Rhþ and two ligands to be the predominant
species. The rest of the ion population includes multiple charged
species, species containing counterions, singly charged species
due to deprotonation, and free uncoordinated ligands. This
mixture is partially the result of the processes that take place
during droplet shrinkage right after electrospray. Due to the
difference in environment provided by the droplet and the gas
phase, multiply charged ions present in the droplets engage in
energetically favorable processes that lead to net charge reduc-
tion: intra- and intercluster Hþ transfers, the transfer of a proton
between two ions, and binding of a counterion.15 These un-
avoidable processes complicate the acquired spectra.
To reduce this complexity, several mixtures of charged ligand 10
and uncharged ligand 1 were studied. After extensive experimenta-
tion it was found that applying a mixture of cationic ligand 10 and
neutral 6-DPPon (1) in a 1:1 ratio in the presence of 1 equiv of the Rh
precursor provided less complex spectra while in the hydroformyla-
tion reaction maintaining high catalytic activity and selectivity.
A closer look into these spectra shows no obvious candidates
for some of the species postulated in the Wilkinson cycle, as
expected because of the absence of synthesis gas, thus prompting
further experiments under hydroformylation conditions.
(14) (a) See ref 8a and: (b) Waloch, C.; Wieland, J.; Keller, M.; Breit,
B. Angew. Chem., Int. Ed. 2007, 46, 3037 for hydroformylation in polar
solvents applying different heteroaromatic platforms.
(15) Kebarle, P.; Yeughaw, H. O. In Electrospray Mass Spectro-
metry; Cole, R. B., Ed.; J. Wiley & Sons: New York, 1997; Chapter 1.

Article
Organometallics, Vol. 29, No. 11, 2010 2527
Scheme 12. CID Experiment with m/z 848
Scheme 13. CID Experiment with m/z 848 under CO/D2
Live Streaming. The next logical step was to live stream reaction
and B in the commonly accepted catalytic cycle proposed by
Wilkinson,16 Heck, and Breslow,17 as shown in Scheme 7.
Without substrate present in the reaction mixture, several ions
containing CO and H2 could be detected, m/z 848, m/z 876, and
the “resting state” m/z 1127. In addition, the corresponding com-
plexes derived from two charged ligands are present also (m/z 503
and 517), as depicted in Scheme 8. The assignment of the mole-
cular structures is based on CID experiments, i.e., collision of the
parent ion with an inert gas at different collision energies and
monitoring of the resulting fragments (daughter ions).
media into the mass spectrometer using the techniques developed
for that end. The above-mentioned mixture was subjected to a CO/
H2 atmosphere (20 bar) in a stainless steel autoclave equipped with
a capillary connected directly to the ESI-MS. The pressure inside
the autoclave pushes the analyte mixture all the way through the
pipe to the electrospray tip of the MS, and a pair of split valves
regulate the flow, as described elsewhere.3
Under these conditions several new complexes are antici-
pated to be present corresponding to the intermediates A, A0,
(16) Evans, D.; Osborn, J. A.; Wilkinson, G. J. Chem. Soc. A 1968,
3133.
(17) (a) Heck, R. F.; Breslow, D. S. J. Am. Chem. Soc. 1961, 83, 4023.
(b) Heck, R. F.; Breslow, D. S. J. Am. Chem. Soc. 1962, 84, 2499.

2528 Organometallics, Vol. 29, No. 11, 2010
Scheme 14. CID Experiment with m/z 848 and CO As Collision Gas
Beierlein et al.
Scheme 15. Alternative Structure for B (right side)
The structure of the resting state (m/z 1127), which has been
described with 6-DPPon before and affirmed by X-ray analysis,
was tested in CID experiments. Using argon as collision gas this
ion yields several fragments, an ion with m/z 848 resulting from
the loss of one 6-DPPon ligand and m/z 437, which corresponds
to 10, thus suggesting the assumed structure A0 (Scheme 10).
To our surprise, a very prominent ion of m/z 818, which
corresponds to [Rh(6-DPPon)(10)(-H)]þ, is always present.
This ion has not been previously spotted using other techniques.
A complex that might be comparable in structure could be
prepared upon reaction of 6-DPPon and [Rh(COD)acac].
X-ray analysis revealed the structure depicted in Scheme 11.
One pyridone has lost a proton, resulting in a strong triden-
tate P,N,P coordination motif. Such a coordination mode could
also be present in the detected ion at m/z 818, making it more
sterically compromised, or even occupying all coordination
sites, rendering it unreactive.
In addition to the analytical detection of solution-phase inter-
mediates, ESI-MS can provide information about the structure
and reactivity of these intermediates by conducting gas-phase
ion-molecule reactions of the species of interest with suitable
reactants.1,2
When reacting the ion at m/z 818 with Ar, D2, CO, olefins, or
aldehydes in the gas-phase at different collision energies and

Article
Organometallics, Vol. 29, No. 11, 2010 2529
Scheme 16. CID with m/z 960
pressures, no new signals could be detected other than the
fragments already mentioned.
Scheme 17. Possible Complex E0 and Fragmentation Pattern of
m/z 960
Furthermore, the ion at m/z 848 (B), which represents the
starting point of the catalytic cycle and is normally not detect-
able in solution, was of special interest and investigated further
by means of CID experiments in the gas phase.
Upon CID, this ion loses 30 mass units to produce an ion of
m/z 818 or loses 511 m/z units to produce an ion of m/z 437 that
corresponds to 10 (see Scheme 12).
Assignment of this species (the hydridocarbonyl complex)
is confirmed by an experiment using D2 instead of H2. The
mass shift of the corresponding species B-d1 is þ1 (Scheme 13).
Consequently the fragmentation pathway leading to the loss of
CO and H2 becomes the loss of CO and HD by CID of B-d1. The
mechanistic interpretation of this fragmentation will be dis-
cussed in detail below.
By those observations we concluded that the structure of the ion
at m/z 848 is indeed B in the cycle above (Scheme 7). To get more
information about the true nature of the complex, it was reacted
with CO as collision gas, producing an ion at m/z 876 in very low
yield and only at low collision energies (0 to 6 eV (lab scale),
Scheme 14) that is also spotted when synthesis gas is present and
represents A (vide supra). No further fragments are created at
higher energies. Upon reaction with olefins, no adducts are
detected and fragmentation seems to be the only pathway.
As the unsaturated complex B should have a square-planar
geometry with free coordination sites, this lack of reactivity
seems unlikely. Bearing the observed loss of H2 in mind, one
could envision a modified structure of B that results from an
oxidative addition of the N-H part of the hydrogen-bond
framework. This indeed could be the case here, as some pre-
liminary DFT calculations of the parent system complex
[HRh(CO)(6-DPPon)2] indicate that the OH-O bond is much
stronger than the NH-N bond, the latter opening up to a great
extent in the course of the catalytic cycle, thus making it
accessible for the proposed oxidative addition (Scheme 15).
In solution, as depicted on the left side of Scheme 15, the
species B would react immediately with either additional ligand
or a CO molecule, forming the saturated species m/z 876.
However, this pathway is not accessible in the gas phase, as no
uncharged reaction partners are present. The assumed structure
B0 indeed could explain the observed lack in reactivity when
using olefins as collision gas, as no free coordination sites are
present. Syn eliminiation of H2 would be feasible from B0,
followed by loss of CO (or the other way around), giving the
complex with m/z 818, which is observed regardless of the kind
of collision gas used.
Reactions with Substrate in the Mixture. After the successful
identification of crucial intermediates of the catalytic cycle (A,
A0, B), the next logical step was to analyze a mixture under bona
fide hydroformylation conditions, i.e., in the presence of an
olefinic substrate. We decided to use 1-hexene, as it reacts quite
readily under the applied conditions.
In first experiments, we applied standard hydroformylation
conditions ([Rh]:L1:L2:1-hexene = 1:1.1:1.1:100, CO/H2 =
1:1, 20 bar, CH3CN, c(1-hexene) = 0.36 M). However, no
new ions could be detected. Hence, either the steps in the
catalytic cycle are faster than our time frame or the concentra-
tions of the species of interest lie below the detection limit of the
method. Indeed, when increasing the amount of substrate to a
concentration of 50 vol % of 1-hexene, a possible acyl complex
at m/z 960 (E or F) could be detected and further investigated.
Additionally, an ion of m/z 932 (C or D) was present too, but the
low intensity of the signal disallowed CID experiments.
Upon CID with Ar or D2, the ion at m/z 960 reveals several
interesting fragments, namely, m/z 848, which could be B in the
catalytic cycle, and m/z 818. No ions such as m/z 932 (C or D),

2530 Organometallics, Vol. 29, No. 11, 2010
Scheme 18. ESI-MS Spectra under CO/D2 Atmosphere and with Substrate in Solutiona
Beierlein et al.
a Highlighted areas are shown in Schemes 19 and 20.
Scheme 19. ESI-MS Spectra under CO/D2 Atmosphere and with Substrate in Solution (m/z 750 to 860)
where one CO is lost, or m/z 876 (A), where only the unreacted
olefin is lost, were detected (Scheme 16).
failed to do so in our experiments. Furthermore, it reacts in the
same manner, giving m/z 848 with Ar as collision gas, where
only fragmentation would be expected. One logical conclusion
for this unexpected reactivity is that the structure of m/z 960
resembles neither the proposed complex E nor F. Instead, one
could think of the olefin complex E0 resulting from β-hydride
elimination (see Scheme 17).
An ion equivalent to m/z 960 but at m/z 980 is present when an
excess of styrene is used instead of 1-hexene, but its signal proved
too weak for further CID experiments. Styrene is thought to
form a stable acyl complex with Rh under hydroformylation
conditions, proved by NMR studies.18
The structure of m/z 960, which should correspond to E or F,
should react with D2 to regenerate A þ 1 m/z unit. However, it
Colliding E0 with any collision gas would result in the loss of
the weakly bound olefin ligand, giving m/z 876, which is A in
Scheme 7. As shown before (Scheme 14), this complex is stable
only under very low collision energies, instantaneously losing
the second CO molecule, giving m/z 848. Alternatively, a direct
(18) Brown, J. M.; Kent, A. G. J. Chem. Soc., Perkin Trans. 2 1987,
1597.

Article
Organometallics, Vol. 29, No. 11, 2010 2531
Scheme 20. ESI-MS Spectra under CO/D2 Atmosphere and with Substrate in Solution (m/z 430 to 450) after 20 min (left) and 2 h (right)
Reaction Time
Scheme 21. Proposed Pathway for the D2 Incorporation in 10
β-hydride elimination from F, liberating a ketene, would result
directly in the complex m/z 848 (Scheme 17, left side). Although
a reaction like this seems unlikely and has not been reported
for Rh complexes, there is evidence for the elimination of
a ketene from comparable Ru complexes [Ru(COMe)(CO)2-
(triphos)]Cl.19
D2-Incorporation Experiments. Using D2/CO as synthesis gas
gave more insight into the nature of the ion at m/z 848 and the
catalytic cycle itself. When no substrate is present, only one
deuterium atom is incorporated to B, shifting its mass up one
m/z unit. Upon collision, 31 or 30 m/z mass units were lost.
Incorporation of deuterium was not complete, and thus only a
part of the ion population at m/z 849 had a deuterium atom
bound to the Rh center. The ion corresponding to 10 was
detected at m/z 437 and 438 (see above).
Adding a substrate proved revealing. While the main mass
shift for B was 2 mass units (Scheme 18) and upon collision the
same behavior as before was seen, the free ligand was detected as
a mixture of ions with m/z 437 and 438 (Scheme 20).
This finding clearly points to the involvement of the H atoms
of the hydrogen-bond framework into the catalytic cycle, as
turnover seems to be crucial for the observed incorporation of
deuterium into the ligand. This incorporation is progressive, as
the right spectrum in Scheme 20 was taken two hours after the
start of the reaction, reaching approximately a 2:1 ratio for
heavy versus normal.
To explain the observed deuteration, two different pathways
seem to be reasonable; the first one, being more closely related to
the commonly accepted catalytic cycle, is depicted in Scheme 21.
After loss of CO and H2 a complex with P,N,P or even double
P,N coordination results that partially resembles the structure
being identified by X-ray (see Scheme 11). The close and stable
P,N coordination might be the reason for the lack of reactivity in
our gas-phase investigations. In solution, on the other hand, due
to the presence of a vast variety of competing coordinating
agents (CO, solvent molecules, abundant ligand), one might
envision a weaker bonding of the ligands. Activation of deuter-
ium or hydrogen can take place over a thermodynamically
preferred six-membered transition state, resulting in the incor-
poration of deuterium into the ligands, as observed. CO co-
ordination regenerates in the last step the now deuterated
complex 14.
The second pathway we like to propose is based on the above-
mentioned oxidative addition step possible for complex B, giving
rise to a new mechanism (Scheme 22).
The intermediate F could undergo an oxidative addition yield-
ing the Rh(III) complex F0, followed by reductive elimination to
(19) Hommeltoft, S. I.; Baird, M. C. J. Am. Chem. Soc. 1985, 107,
2548.

2532 Organometallics, Vol. 29, No. 11, 2010
Scheme 22. Proposed New Hydroformylation Mechanism for Self-Assembling Ligands
Beierlein et al.
liberate the product aldehyde. The resulting complex 15 can be
seen as structurally related to 13 in Scheme 21, having one more
CO ligand. H2/D2 activation can proceed also over a six-mem-
bered transition state (16), regenerating the now deuterated
complex 14, which enters a new catalytic cycle. This unprece-
dented catalytic cycle would not only explain the observed in-
corporation of deuterium into the free ligand but explain why we
never detected a complex G that would be the result of oxidative
addition of hydrogen: if the hydroformylation reaction applying
our self-assembling ligands could indeed proceed in the above
depicted way, a complex G with m/z 962 simply would not occur.
Nonetheless, at this point we cannot comment on the possi-
bility of these mechanistic pathways being by any means part of
the catalytic cycle of the hydroformylation itself in solution and
not only an orthogonal equilibrium phenomenon. The unique
nature of our self-assembling ligand systems and the highly
dynamic tautomeric behavior of the applied 6-DPPon ligand
could be the basis for this unprecedented reactivity. These
findings seem to be in accordance with literature examples where
a similar involvement of two hydrogen atoms;one at the metal
center, another at a heteroatom;occurs.20,21
Further investigations, especially with our other heterobiden-
tate self-assembling ligands, together with quantum mechanical
calculations to further affirm the observed phenomena, are on-
going and will help to shed more light on the mechanism of the
hydroformylation and other transition-metal-catalyzed reactions.
Acknowledgment. This work was supported by the
DFG, the International Research Training Group “Cat-
alysts and Catalytic Reactions for Organic Synthesis”
(IRTG 1038), and the Alfried Krupp Award for young
€
university teachers (to B.B.). We thank Gunther Leon-
hardt-Lutterbeck for great technical assistance.
(20) Casey, C. P.; Singer, S. W.; Powell, D. R.; Hayashi, R. K.;
Kavana, M. J. Am. Chem. Soc. 2001, 123, 1090, and references therein.
(21) Yamakawa, M.; Ito, H.; Noyori, R. J. Am. Chem. Soc. 2000, 122,
1466.
Supporting Information Available: This material is available
free of charge via the Internet at http://pubs.acs.org.