Full Paper
vidual effects of RZnBr and ZnBr2. Nonetheless, the inclusion of
heterobimetallic Pd–Zn complexes is essential for the success-
ful modeling of the kinetic measurements.
tected [L2PdZnBu]+ both in the absence and presence of LiBr
and the comparison of absolute or relative signal intensities
between different samples is notoriously difficult.
In addition to the implications discussed above, the pres-
ence of heterobimetallic Pd–Zn species in Negishi cross-cou-
pling reactions raises further questions. As these complexes
contain both low-valent Pd and an organozinc moiety, the oxi-
dative addition and transmetallation do not need to proceed
in a consecutive manner, but could possibly also occur in a con-
certed fashion. Alternative pathways could reverse the order of
the elementary steps and start with the transmetallation, fol-
lowed by the oxidative addition. Terao and Kambe have
suggested mechanisms of this type in a related context.[40] Our
gas-phase fragmentation experiments demonstrate that the or-
ganyl group R of the [L2PdZnR]+ complexes can indeed be
transferred to the Pd center [Eqs. (5c) and (5d)]. However, the
latter adopts an oxidation state of +II in the course of this re-
action and, thus, would no longer be prone toward oxidative
addition.
As an alternative rationalization for the effect of LiBr, we also
consider the possible involvement of Pd0 ate complexes in the
present experiments.[20] On the one hand, the coordination of
a Brꢂ ion to an [LPd] complex enhances the electron density of
the latter, thus possibly also increasing its affinity toward
Lewis-acidic RZn+ or RZnBr species. On the other hand, the
concomitant saturation of a vacant coordination site on Pd0
supposedly lowers its tendency toward further association re-
actions. If the second effect prevails, the Pd0 ate complexes
could also counteract the formation of Pd–Zn dimers.
However, our experiments do not provide any direct evidence
for the presence of Pd0 ate complexes.
In addition to its rate-enhancing effect, LiBr also helps to
protect the catalyst from irreversible deactivation. We will
address this issue in the next section.
Catalyst aggregation and irreversible degradation
Effect of LiX additives
As 31P NMR spectroscopy has demonstrated, the catalytic
system under investigation is of considerable complexity al-
ready in the absence of any added reagents. The formation of
several phosphine-containing products is surprising given the
strong tendency of S-PHOS to afford only mono-ligated Pd0
complexes.[27,29] Possibly, different conformers[41] and/or differ-
ent aggregation states are present. Moreover, the tempera-
ture-dependent 31P NMR spectroscopic measurements point to
the operation of dynamic equilibria involving free S-PHOS. The
latter should not remain if the PdII precatalyst were completely
reduced to Pd0 and if exclusively complexes of 1:1 Pd/S-PHOS
stoichiometry formed. Possibly, the reduction of Pd(OAc)2 did
not reach completion and/or produced larger Pd aggregates,
which were less ligated on average. The presence of larger ag-
gregates was directly suggested by the absorption over an ex-
tended range of wavelengths observed in the UV/Vis spectra
at longer times; such behavior is typical of Pd (and other
metal) nanoclusters.[42] Similarly, the darkening upon addition
of the organozinc reagent indicates the formation of Pd black,
which necessarily involves Pd aggregates as intermediates and
requires the initial presence of residual, unreduced Pd(OAc)2.
Additional information is afforded by the ESI-mass spectro-
metric observation of [L2Pd2X]+ complexes (X=Br, I, OAc,
PCy2). Related monocationic Pd dimers not only have been ob-
served in previous ESI mass-spectrometric studies,[43,44] but
have also been found in the condensed phase and structurally
characterized.[43] For the very case of L=S-PHOS, Barder pre-
For all reactions investigated, the addition of LiBr (used in
combination with RZnBr) led to a boost in reactivity. Unlike the
situation in more polar solvent mixtures,[15b,c] this increase in
reactivity was brought about by just one equivalent of LiBr,
whereas the addition of further LiBr had no significant effect
and therefore excludes the involvement of so-called higher-
order organozincates, such as RZnBr3 or LiRZnBr3ꢂ. LiCl,
2ꢂ
LiClO4, and LiI (used in combination with BuZnCl and BuZnI,
respectively) also accelerated the reaction.
For Pd-catalyzed Negishi cross-coupling reactions in solvents
more polar than THF, the observed rate-enhancing effect of LiX
additives was rationalized by the formation of organozincates,
the increased nucleophilicity of which supposedly accelerates
the transmetallation step. We[16] and others[17–19] indeed have
established the tendency of RZnX reagents (and ZnX2) to react
with lithium halides and form the corresponding zincate
anions; the present negative-ion mode ESI mass spectra were
also dominated by zincate anions most of which were purely
inorganic, however (the absence or reduced abundance of or-
ganozincates observed in the present work presumably origi-
nated from an imperfect insulation of the ESI source from the
atmosphere and the occurrence of hydrolysis reactions of
these sensitive species). As our kinetic measurements have
identified the oxidative addition, and not the transmetallation
as the rate-determining step, the given explanation cannot
apply to the present case. However, in an indirect way, the
organozincate complexes may also act upon the oxidative ad-
dition. In comparison with neutral RZnX, the corresponding or-
ganozincate anions have a drastically reduced Lewis acidity
and, thus, presumably do not easily combine with electron-rich
Pd0 to yield heterobimetallic Pd–Zn complexes. The (partial)
persistence of free [LPd] would then account for the increased
reactivity in the presence of LiBr (and other lithium halides).
Direct experimental proof for this explanation is not straight-
forward because the ESI mass-spectrometric measurements de-
ꢂ
pared [L2Pd2]2+(BF4 )2 by treatment of [L2PdCl2] with AgBF4.[45]
X-ray crystallography revealed a direct Pd–Pd bond and an
end-on coordination of the two phosphine ligands, which fur-
ther interact with the Pd centers by their aryl groups.[45] A very
similar geometry may be assumed for the monocations detect-
ed in the present study, in which the additional anionic ligand
could adopt a bridging binding mode. Like their well-known
neutral counterparts,[46] the cationic Pd dimers are commonly
thought to be composed of two covalently bound PdI centers
Chem. Eur. J. 2015, 21, 1 – 14
9
ꢂ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ