Oxygen-Transfer Reactions
FULL PAPER
formed on these optimized structures with a larger (triple-z)
6-311+GACTHNUTRGNEUG(N d,p) basis set on C, H, O and Cl. Where it was
computationally tractable, we also performed optimizations
at the “double-hybrid” density functional level of theory
with the B2PLYP functional.[32] This method replaces a frac-
tion of the semilocal correlation energy by a nonlocal corre-
lation energy expression that employs the Kohn–Sham orbi-
tals in second order perturbation theory and delivers im-
proved energetics over hybrid density functionals such as
B3LYP. Geometries and energetics obtained at this level
and from single-point energy calculations at the SCS-MP2
level[33] were also compared with the B3LYP and M06-2X
results, to verify that the hybrid-GGA functionals gave ge-
AHCTUNGTREGoNNNU metries and energetics agreed with these more demanding
methods. Transition structures were confirmed by the pres-
ence of an imaginary harmonic vibrational frequency corre-
sponding to a displacement along the proposed reaction co-
ordinate. NBO version 3.1 was used for the calculation of
Wiberg bond order (BO) from the natural bond orbitals.[34]
Competing [2+2] and [4+2] reaction coordinates were
computed for a model substrate for which R1, R2, R3, and
R4 are all methyl groups, allowing us to compare the perfor-
mance of DFT at the hybrid-GGA level with more demand-
ing B2PLYP and SCS-MP2 calculations. Experimentally, we
demonstrated that similar substrates (Table 1, entry 7: R1,
R2, R4 =Me/R3 =Ph and entry 14: R2, R3, R4 =Me/R1 =Ph)
rearranged in 91–92% yield. We begin with a discussion of
our newly proposed [4+2] pathway, for which the computed
B2PLYP reaction coordinate is shown in Figure 1. The rate-
limiting step in the rearrangement is the intramolecular nu-
cleophilic addition to the Au-coordinated alkyne—the acti-
vation barrier for this step is computed to be a modest
8.8 kcalmolÀ1, forming five-membered ring oxonium inter-
mediate B (via 5-endo-dig TS-1). Once this cyclization has
occurred to form C the remaining transformations that lead
to the rearranged product are all computed to be relatively
facile. The transition state TS-2, formally a [4+2] cycloaddi-
Scheme 6. Synthesis of substrate 1o-18O and its 13C NMR spectrum.
Scheme 7. 18O isotopic experiment of alkynylketoester and its 13C NMR
spectrum.
products (2b-18O and 2o-18O) demonstrates that the [2+2]
pathway is disfavored, and instead it is the newly proposed
[4+2] pathway that accounts for the mechanism of the
gold-catalyzed, intramolecular, oxygen transfer of 2-alkynyl-
1,5-diketones and 2-alkynyl-5-ketoesters.
À
À
tion, involving the formation of two new C C and C O s-
bonds lies only 4.4 kcalmolÀ1 above the starting complex 1.
The barriers for TS-3, the opening of acetal intermediate C,
and TS-4, the opening of oxonium D, are very small indeed.
At room temperature this process would be expected to
occur readily, consistent with the 5 min reaction times ob-
served experimentally.
Alongside our experimental studies, we turned to quan-
tum chemical calculations to seek further verification that
the proposed [4+2] mechanism is the preferred pathway
and to understand the origins of this selectivity. We also
sought to clarify the mechanism of the competing [2+2]
pathway, which has previously been invoked on a number of
occasions to rationalize carbonyl–alkyne metathesis reac-
tions. All calculations were performed with Gaussian 09.[27]
Stationary points were fully optimized with the B3LYP[28]
(hybrid GGA) and M06-2X[29] (hybrid meta-GGA) density
functionals, using a fine grid for numerical integration. Both
functionals have been utilized extensively in computational
studies of AuI catalysis,[30] although the M06-2X functional
may be expected to describe nonbonding interactions with
greater accuracy. For optimizations we used a combination
of the Pople 6-31G(d) basis set for C, H, O and Cl and the
LANL2DZ (Hay–Wadt) basis including an effective core
potential for Au.[31] Single-point calculations were per-
B2PYLP optimized transition structures (TSs) along the
[4+2] pathway are also shown in Figure 1. Rate-limiting
TS-1 involves a 5-endo-dig cyclization, which is allowed ac-
cording to Baldwinꢁs rules. Bond formation in TS-2 can be
À
seen to be highly asynchronous, with the forming C O bond
À
at 1.66 ꢂ (Wiberg BO 0.60) and the C C bond at 2.67 ꢂ
(Wiberg BO 0.12); however, this process is concerted as no
intermediate exists between B and C. All of the computa-
tional levels examined suggest the rate-limiting step in the
[4+2] pathway is carbonyl-oxygen addition to the Au-coor-
dinated alkyne, giving energetic profiles similar to those
shown in Figure 1 and optimizations with either B2PLYP,
B3LYP or M06-2X density functionals result in similar ge-
Chem. Eur. J. 2011, 17, 10690 – 10699
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
10693