Organic Letters
Letter
as synthetic intermediates has resulted in considerable effort to
develop synthetic methodology to access them.14 These
methodologies often rely on the Corey−Chaykovsky reac-
tion15 or the use of functionalized diazo compounds.16
A survey of the literature revealed that the established
methods described above would be unsuited to this purpose,
and indeed, this proved true in practice. We therefore decided
that an alternative to C−C coupling was required, as well as
allowing the use of more readily available starting materials. To
this end, we have developed the first hydrosilylation reaction of
propargylic amines to afford a single regioisomer in high yields.
Platinum-catalyzed hydrosilylation reactions were first
described in the 1950s,17 with the disclosure of hexachlor-
oplatinic acid.18 Since then, the field has matured, with the
development of more sophisticated catalytic systems being
disclosed.19 Of these, platinum is still widely used, due to its π-
Lewis acidity and functional group tolerance. As a result of the
increased cocoordinative ability of nitrogen, platinum-catalyzed
hydrosilylation of amine-containing moieties is scarcely
reported in the literature. Speier disclosed the attempted
hydrosilylation of allyl amines, resulting in the formation of an
isomeric mixture of silanes (Figure 1).20
Given that the poisoning of hexachloroplatinic acid by amine
and amide moieties is documented, this is not particularly
surprising. Endo demonstrated the relationship between the
ratio of nitrogen donors to platinum centers and the
retardation effect on catalyst efficiency.21 When studying the
hydrosilylation of N-allylamines, Chechelska-Noworyta found
that it was necessary to decrease the nucleophilicity of the
nitrogen through either steric or electronic factors. Without
tuning these factors, the regioselectivity drops to approximately
2:1 β:α.22 Furthermore, Pregosin and co-workers have shown
that both aliphatic and aromatic amines preferentially displace
olefinic ligands in the complex formed between PtCl2 and
styrene.23 These investigations show that this transformation is
nontrivial in nature and represents an unmet need within the
synthetic community.24 As mentioned, we required a robust
method to produce these important allylic amine intermediates
with a range of pendant functional groups. 1a was chosen as a
model substrate, as it does not deactivate the amine moiety
toward the aforementioned metal complexation.
We began our investigation by treating 1a with dimethyl-
phenylsilane in the presence of PtCl2 as catalyst (Figure 2).
Unsurprisingly, this provided an inseparable mixture of
products, with the desired β-isomer being formed preferen-
tially. We then began a survey of monodentate phosphine
ligands, which improved the selectivity profile of the reaction.
Although improved, the undesired α isomer was still produced
in non-negligible quantities, with the best result observed when
using XPhos in a 1:2 ratio relative to the catalyst.25 We
reasoned that the monodentate ligand was being displaced by
the amine during the reaction, and reducing the impact of the
bulky ligands on selectivity. We therefore switched our focus to
bidentate ligands and found that Xantphos (Figure 1, entry 8)
was unique in its ability to provide full conversion and
complete control (>99:1) of regiochemistry, providing the
desired β-isomer as the sole product. Further exploration of
solvent, temperature, and catalyst loading resulted in a less
efficient reaction.
regioisomer. Symmetrical (2e) and unsymmetrical propargyl
amines (2c, 2d, 2f) also worked well, producing allyl amines
with multiple points of derivatization. The reaction also
tolerated diverse electronics, with electron-donating (2g, 2h)
and -withdrawing (2i) as well as mixed (2j) aryl groups all
working well. Furthermore, electron-withdrawing (2k) and
-donating amines (2l) proceeded smoothly.
As we envisaged this methodology to be used in in-house
medicinal chemistry programs, we wanted to explore the effect
that sulfonamides, amides, and carbamates had on the reaction.
We were initially concerned that the extra points of interaction
would compete for the active metal species; however, our
concerns were, thankfully, misplaced.
As shown in Figure 4, sulfonamides bearing weakly (4b) and
strongly (4c) electron-donating and electron-withdrawing
groups (4c, 4d) were all well tolerated, producing fragments
with multiple functional handles. Mixed electronics (4f) were
also successful in the reaction, providing the allyl amine in high
yield. In a similar vein, amides also worked very well, with the
vinyl silanes produced in excellent yield throughout. Aliphatic
(4g), aryl (4h), and heteroaromatic (4i, 4j) amides all
proceeded in a straightforward manner, affording a single
regioisomer in each case. Finally, we explored the use of well-
established carbamate protecting groups. Both Boc (4k) and
Fmoc (4l) proceeded smoothly, providing scaffolds with
complementary deprotection strategies.
As noted, ready access to multifunctional aziridines was a
key goal in this research. We reasoned that an alternative
approach to these heterocycles could be realized, using the
well-established β-silicon effect to our advantage.26 In
particular, we wanted to explore if an iodination triggered
ring closure could be used to access ambiphilic substrates
(Scheme 2).
We therefore subjected 4a to N-Iodosuccinamide (NIS) at
room temperature, and saw complete conversion to the
product after 12 h. Delighted that our hypothesis was proved
correct, we explored the reaction further. As shown (Figure 5),
a range of aziridines can be produced in high yields including
electron-donating (5a−c) and -withdrawing (5d, 5e) as well as
mixed (5f) electronics. When 2a or 4g was subjected to same
reaction conditions, an inseparable mixture of products was
formed, presumably due to competing migrations and
alternative cyclization modes as reported by Taguchi.27
Pleasingly, when 4x was treated with N-Bromosuccinamide
(NBS), the halogenation−aziridination reaction was also
effective, affording 5g in 68% yield.
Finally, we wanted to explore if the aziridine product could
be obtained in a telescoped fashion (Scheme 3).28 This has the
benefit of reducing the number of chromatography operations,
as well as increasing the overall economy and efficiency of the
process. To this end, we subjected 3e to the hydrosilylation
conditions followed by NIS. This strategy proved successful,
and the aziridine product (6e) was obtained in 73% over the
two steps.
Based on previous studies of the reactivity of platinum−
phosphine complexes, in addition to the highly studied Chalk−
Harrod hydrosilylation mechanism25a and the stereo- and
regiochemical outcome of the transformation, we propose the
following reaction mechanism (Scheme 4). Initially, Pt(II)
complex (I) is formed via complexation of XantPhos to PtCl2
to form cis-Pt(XantPhos)Cl2. This complex has been
previously isolated and fully characterized, and the 31P NMR
spectra of the complex formed under our reaction conditions
With these conditions in hand, we turned our attention to
probing the substrate scope of the reaction (Figure 3). The
reaction was tolerant to secondary (2a, 2b, 2j, 2k) propargyl
amines, providing the desired product in high yield as a single
4465
Org. Lett. 2021, 23, 4463−4467