60
A. Bruneau-Voisine et al. / Journal of Catalysis 347 (2017) 57–62
Scheme 5. Principal bond lengths (Å) of complexes 4 and 5 (in blue and red the difference in between the bond length of 4 and 5).
also be methylated in satisfactory yield. It must be noted that an
acidic phenol substituent inhibited the reaction (b14).
Furthermore, sulfonamides (a20–23), which are common moi-
eties in biologically active compounds [102], could also be methy-
lated with this methodology with excellent yields, although under
harsher conditions (1.2 equiv. of base for 60 h) [42].
Finally, in a Young type NMR tube, a mixture of 4, t-BuOK
(2 equiv.) and methanol (5 equiv.) in toluene-d8 was heated at
120 °C for 18 h. The 31P{1H} NMR displayed three broad singlets,
at 164.2 ppm, 132.9 ppm and 129.5 ppm, matching with the 31P
chemical shift of hydride 3 [79], complex 4 (or the neutral
bromo-dicarbonyl manganese complex [101]) and 5, respectively
(see E.S.I.). The presence of the hydride was also ascertained by
the presence of the characteristic triplet at ꢀ5.8 ppm in the 1H
NMR [79]. The same signals in 31P NMR were also observed when
a catalytic reaction, i.e. in the presence of aniline, was carried out
in an NMR tube (see E.S.I.). These experiments show that upon
heating, in the presence of methanol and base, the manganese
hydride 3 is formed. Finally, the deprotonated complex 5 was
tested as catalyst without the addition of external base under stan-
dard conditions: with 10 mol% of 5, the N-methylaniline was
obtained in 30% yield, showing that an excess of base is needed
to reactivate in situ the catalytic species.
Finally, starting from 1,8-diaminonaphthalene, 1H-perimidine
(b24) was obtained as the major product, resulting from the
cyclization of the imine intermediate to the aminal derivative
which undergoes dehydrogenation to yield the cyclic formamidine
[103] (see Scheme 3).
Next, we performed a series of experiments to get mechanistic
insights. Similar catalytic systems, based on PN3P pincer ligands,
have been previously studied by Huang et al. with ruthenium
and iridium complexes [104–107], and DFT calculations have been
performed by Veiros et al. [79] on complex 3. Based on their
results and on our calculations on rhenium PNP complex [108],
we assumed that in the presence of a base and methanol: (i) the
N–H of pre-catalyst 4 should be deprotonated inducing the
de-aromatization of the pyridine ring and facilitating the decoor-
dination of one CO ligand to produce the active catalyst as a neu-
tral dicarbonyl 16 electrons manganese complex and (ii) the
neutral 16 electrons complex should after dehydrogenate
4. Conclusions
In conclusion, we have demonstrated that PN3P manganese pre-
catalyst efficiently catalyzes the selective methylation of aniline
derivatives with methanol under hydrogen borrowing conditions.
Compared to our previous system, the reaction is now operating
under a catalytic amount of base. The scope of the functional group
tolerance is large, making this approach appealing for synthesis.
Furthermore, we have been able to isolate and characterize by X-
ray diffraction studies a de-aromatized manganese intermediate,
shedding light on the catalytic mechanism.
methanol to form the manganese hydride complex
formaldehyde.
3 and
Indeed, in toluene, upon the addition of t-BuOK, the solution
turned to blue, leading after addition of pentane to a very sensitive
blue solid that we were not able to characterize further due to its
possible paramagnetic nature. By analogy with the work of Mil-
stein [77], we postulated that this compound was the de-
aromatized di-carbonyl 16 electrons complex. Upon the addition
of methanol to this powder, the solution immediately turned to
yellow. After slow evaporation of the solvent in a glove box, single
crystals suitable for X-ray diffraction studies were grown. Surpris-
ingly, we did not obtain the manganese hydride 3, but a neutral, 18
electrons, de-aromatized manganese complex 5 bearing three car-
bonyl ligands (depicted in Scheme 4). The deprotonated complex is
stabilized in the solid state by a network of hydrogen bonding
though molecules of methanol that have co-crystallized. The
length of the bonds was significantly modified in 5 compared to
4 (Scheme 5), showing the de-aromatization of the pyridine-ring.
Complex 5 was also generated quantitatively at r.t., by reaction
of 4 with 1 equiv. of t-BuOK, in a mixture of toluene and methanol
(1:1, v:v). The 31P{1H} NMR in DMSO-d6 displayed a single signal at
130.18 ppm (d(31P{1H}, tol-d8) = 129.7 ppm), probably due to a fast
scrambling of the N–H protons. In the 1H NMR, the aromatic pro-
tons of 5 appeared as two sets of signals (d = 6.75 (t, Hpara), 5.59
Acknowledgments
We thank the Centre National de la Recherche Scientifique
(CNRS), the Université de Rennes 1, Fonds Européens de
développement économique régional (FEDER founds), INCREASE
FR-CNRS 3707, la Fondation Rennes 1 (grant to D.W.) and the
Agence National de la Recherche (ANR agency, program JCJC
ANR-15-CE07-0001 ‘‘Ferracycles”).
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
References
(d, 2Hmeta)), shielding compared to the cationic complex
4
(d = 7.48 (t, Hpara), 6.39 (d, 2Hmeta)), which is in line with the
increased electron density of the pyridinyl ring.