Organic Letters
Letter
counterpart A− (Figure 1). Ideally, the desired pKa of HA
would fall within the green region of the diagram due to the
the acetate reagents 1c−1f were formed in ethanol and
precipitated using ether typically on 2−10 mmol scale. The
pivalate reagent 1b, however, requires the use of MTBE and is
also different in other aspects: this anion was chosen due to its
slightly higher basicity (pKa(DMSO) is 12.917a) and its bigger
steric size. As anticipated, 1b, bearing a large tert-BuCOO−
anion more closely matching the cation’s size, is non-
hygroscopic and remains shelf-stable in air while the acetates
1a, 1c, and 1f are hygroscopic and prone to slow hydrolysis by
adventitious moisture. These attractive benefits have to be
weighed, however, against the much higher cost of pivalatess.
Owing to good solubility in a range of solvents, the
olefination of PhCHO with the pivalate 1b works well in
solvents of different polarities (Figure 2), whereby the highest
Figure 1. Side-by-side comparison of basic anions (pKa(DMSO) of
conjugate acids17b,18) and a benzylic QPC [PhCH2PPh3]+.
following: first, if the pKa of HA is too low (bottom red zone),
the anion would be too weak a base to attain a practical overall
Wittig reaction rate. Second, with too basic A− (top red zone,
pKa greatly exceeding that of the reference QPC), in situ
deprotonation reaction would likely lead to isolation of reactive
ylide, defying the goal of having the desired shelf-stable ionic
reagent. Thus, the choice of a suitable basic anion is
significantly narrowed to the green zone.
Figure 2. Eigenbase reagent 1b performs olefination of benzaldehyde
3a in various solvents (conversion to 4a, %).
conversion to stilbene 4a is achieved in THF and its green
alternative 2-MeTHF. Perhaps somewhat surprisingly, even in
acetone an acceptable conversion to stilbene 4a is obtained
indicating the far lower reactivity of ketones.
Our initial efforts were focused on salts of QPC which have
sufficiently high equilibrium acidity14 (e.g., the methylene
group in [PhCH2PPh3]+ has pKa(DMSO) = 17.414a) so that
fluxional acid−base interactions within the QPC-carboxylate
species are likely to generate minor amounts of semistabilized
ylides from, benzylic QPC as the starting point. To our minds,
regarding the choice of anion, its availability and purity was
equally critical. Certain bases, e.g. the imidazolate or phenoxide
anion, can straightforwardly be accessed by reacting MOH (M
= K or Na) with imidazole15 or phenol,16 respectively, but the
presence of free hydroxide or moisture in the product would be
a major concern due to irreversible QPC hydrolysis giving
phosphine oxide. In contrast, the alkali metal carboxylates,
commonly available in dry pure form, are convenient cheap
sources of suitably basic endogenous anions for reagents 1.
Considering that acetic acid has a pKa(DMSO) of 12.3,17 clearly
below the pKa of the [PhCH2PPh3]+ cation, acetates were
chosen for the preparation of new ion-pair reagents for Wittig
olefinations. When a solution of BnPPh3Br in dry ethanol was
treated with an equivalent amount of KOAc and KBr was
separated, the residue produced crystalline QPC acetate 1a
(see Scheme 1) in 99% yield. Although the acetate 1a is a
white crystalline material, its solutions in aprotic solvents, e.g.
DMSO, have a characteristic pale-yellow hue. This behavior
suggests the possibility of fluxional proton transfer leading to
the formation of trace amounts of ylide in solution. The
presence of IP in the system is consistent with the observation
of long-distance through-space interactions by NOE spectros-
The scope of olefination (Table 1) illustrates the efficiency
of benzylic Eigenbase reagents 1a and 1b in reactions with
aromatic, heteroaromatic, and aliphatic aldehydes. Either
acetate 1a (entries 1−8) or pivalate 1b (entries 9−12)
produce alkenes 4a−4k in high yields. As THF affords the best
result, this solvent was used for olefinations with all Eigenbase
reagents (Tables 1 and 2). Overall, electron-withdrawing
groups on the aldehyde facilitate the process (Table 1, entries
2, 5) compared to electron-donating groups (entries 3, 11, 12)
such as ferrocenyl.19 The benefits of the Eigenbase method-
ology are apparent with less common Wittig substrates, e.g.
those containing aromatic hydroxyl groups20 (entries 3, 11),
and enolizable aldehydes (entries 4, 7) where very little
condensation products were observed21 due to the inherently
mild basicity of carboxylates.
The stark difference between the ion-pair reagents and the
alternative procedure using a QPC bromide and an alkali metal
acetate (Scheme 2) clearly shows the benefits of the new
reagents: the reaction of BnPPh3Br with p-anisaldehyde 3b in
the presence of potassium acetate in THF affords only traces of
alkene 4h; in contrast, using Eigenbase reagent 1a, 81% of
alkene 4h was isolated. Thus, we conclude that the acetate
anion has to be present in the form of an IP to be efficient in
the olefination.
A striking example of the expanded olefination scope is of
course the “unfeasible” reaction shown in Scheme 3: owing to
the fluxional behavior of salt 1 and dioxane 5,22 the reaction
leads to cinnamic alcohols in high yields (Table 1, entry 6;
Table 2, entry 3).
The successful isolation of crystalline QPC acetate 1a was
followed by other QPC carboxylates 1b−1f (Scheme 1), which
were prepared in excellent yields (see SI). Analogous to 1a, all
Other olefinations with the o-halogenated benzylphospho-
nium acetate reagent 1c (Table 2, entries 1−4) clearly show
the preparative benefits of the mildly basic ion-pair reagents as
1458
Org. Lett. 2021, 23, 1457−1462