Halide-Dependent Mechanisms of Reductive Elimination from Gold(III)

Article abstract of DOI:10.1021/jacs.5b04613

Two unique organometallic halide series (Ph₃P)Au(4-Me-C₆H₄)(CF₃)(X) and (Cy₃P)Au(4-F-C₆H₄)(CF₃)(X) (X = I, Br, Cl, F) have been synthesized. The PPh₃-supported complexes can undergo both C_aryl-X and C_aryl-CF₃ reductive elimination. Mechanistic studies of thermolysis at 122 °C reveal a dramatic reactivity and kinetic selectivity dependence on halide ligand. For X = I or F, zero-order kinetic behavior is observed, while for X = Cl or Br, kinetic studies implicate product catalysis. The selectivity for C_aryl-CF₃ bond formation increases in the order X = I < Br < Cl < F, with exclusively C_aryl-I bond formation when X = I, and exclusively C_aryl-CF₃ bond formation when X = F. Thermodynamic measurements show that Au(III)-X bond dissociation energies increase in the order X = I < Br < Cl, and that ground state Au(III)-X bond strength ultimately dictates selectivities for C_aryl-X and C_aryl-CF₃ reductive elimination. (Graph Presented).

Full text of DOI:10.1021/jacs.5b04613

Article
pubs.acs.org/JACS
Halide-Dependent Mechanisms of Reductive Elimination from
Gold(III)
Matthew S. Winston, William J. Wolf, and F. Dean Toste*
Department of Chemistry, University of California, Berkeley, California 94720, United States
S
* Supporting Information
ABSTRACT: Two unique organometallic halide series (Ph3P)Au(4-Me-
C6H4)(CF3)(X) and (Cy3P)Au(4-F-C6H4)(CF3)(X) (X = I, Br, Cl, F) have
been synthesized. The PPh3-supported complexes can undergo both Caryl−X
and Caryl−CF3 reductive elimination. Mechanistic studies of thermolysis at
122 °C reveal a dramatic reactivity and kinetic selectivity dependence on
halide ligand. For X = I or F, zero-order kinetic behavior is observed, while
for X = Cl or Br, kinetic studies implicate product catalysis. The selectivity for
Caryl−CF3 bond formation increases in the order X = I < Br < Cl < F, with
exclusively Caryl−I bond formation when X = I, and exclusively Caryl−CF3
bond formation when X = F. Thermodynamic measurements show that
Au(III)−X bond dissociation energies increase in the order X = I < Br < Cl,
and that ground state Au(III)−X bond strength ultimately dictates
selectivities for Caryl−X and Caryl−CF3 reductive elimination.
temperatures (122 °C) (Scheme 1). Although Caryl−I reductive
INTRODUCTION
■
elimination from these complexes is facile, the factors
controlling selectivity of C−X versus C−C bond formation
are unclear due to a lack of other members of the halide family
that could allow a comparative study.
Transition metal-catalyzed transformations proceed through a
series of fundamental steps, i.e., oxidative addition, migratory
insertion, and reductive elimination. To minimize deleterious
side reactions and maximize overall catalyst efficiency, the metal
must undergo the proper series of reactions with excellent
selectivity. A fundamental understanding of the factors that
affect the selectivity of these elementary steps is critical in
designing and improving new metal-catalyzed transformations.
We have recently shown1 that complexes of the type
(Ph3P)Au(aryl) (aryl = 4-F-C6H4, 4-Me-C6H4) undergo a
photochemical oxidative addition to CF3I to give the air- and
moisture-stable Au(III) complexes (Ph3P)Au(CF3)(aryl)(I).
These complexes undergo rapid Caryl−CF3 reductive elimi-
nation when treated with AgSbF6 (Scheme 1). This trans-
In a seminal study, Hartwig has shown that the rates of
reversible Caryl−X reductive elimination from three-coordinate
Pd(II) increase with halide polarizability (X = Cl < Br < I),
while the thermodynamic driving force increases in the order X
= I < Br < Cl.2 However, because C−X (X = halide) reductive
elimination is often endothermic, studies typically rely on using
high-valent late metals such as Cu(III), Pd(IV) and Pt(IV) to
establish a thermodynamic driving force.3,4 In this vein, Au-
catalyzed halogenations likely involve C(sp2)−X reductive
elimination from Au(III),5 and C(sp3)−F6 and C(sp3)−F7
eliminations from Au(III) have also been demonstrated.
Caryl−X reductive elimination is not necessarily productive,
and may be a decomposition pathway for high-valent
organometallic species with halide ligands. Importantly, Au(III)
catalysts, which are often generated using dihalogen (or formal
dihalogen) oxidants and stabilized by halide ligands,8 could
undergo deleterious, irreversible Caryl−X bond formation to
deplete active catalyst concentrations. With access to a full
family of Au(III) halides, trends in the rates of Caryl−X
reductive elimination from Au(III) could be established.
Perhaps slower Caryl−X bond formation could also be exploited
to promote selectivity for otherwise challenging reductive
eliminations, such as Caryl−CF3 bond formation in complexes of
the type (R3P)Au(aryl)(CF3)(X). Indeed, studies of competitive
Scheme 1. Divergent Reductive Elimination Behavior of
Au(III) Complexes
formation presumably proceeds via the cation [(Ph3P)Au-
(aryl)(CF3)]+. Although this step demonstrates the oxidizing
ability of Au(III) cations, a reliance on stoichiometric Ag(I)
salts to generate the reactive cation is ultimately impractical if a
catalytic process involving such Au(III) intermediates is to be
realized. Due to our failed efforts to induce iodide dissociation
either photochemically or with Lewis acids, we also investigated
thermolytic routes,1 and found that neutral (Ph3P)Au(aryl)-
(CF3)(I) underwent solely Caryl−I reductive elimination at high
Received: May 3, 2015
© XXXX American Chemical Society
A
DOI: 10.1021/jacs.5b04613
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
reductive eliminations should inform factors dictating selectivity
in catalytic cycles.4c,k
trifluoromethylnaphthalene (19F δ: −59 ppm) as an internal
standard. Due to irreversible formation of a new Au(III) species
upon treatment with [Bu4N]X (presumably the aurates
[Bu4N][Au(aryl)(CF3)(X)2] (19F NMR singlet at δ −21 to
−25 ppm), the kinetic order of halide anions could not be
determined. Reactions run in the significantly more polar9
PhNO2 were only slightly affected (see Supporting Informa-
tion), providing evidence against an ionic mechanism involving
tight or dissociated ion pairs.
Herein, we report the synthesis and characterization of a
series of well-defined complexes of the type (R3P)Au(aryl)-
(CF3)(X) (X = I, Br, Cl, F) that undergo both Caryl−X and
Caryl−CF3 reductive elimination with different, halide-depend-
ent kinetic ratios. These ratios vary systematically among the
halide series, showing that halide ligands, often considered
spectators, can dramatically influence reaction behavior.
Thermolysis of Au(III)−Iodide 1-I. As previously reported,
complex 1-I underwent thermolysis at 122 °C to exclusively
generate Ph3PAuCF3 and 4-Me-C6H4−I (t1/2 = 2.5 min).
Consumption of 1-I followed unusual zero-order kinetics over a
range of concentrations (kobs = 4.5 × 10−5 M s−1 from 6 to 35
mM [1-I], Figure 2 and Supporting Information Figure S1).
When 0.005 equiv PPh3 (70 μM) was added, the rate slowed
substantially (t1/2 = 28 min), and the reaction exhibited first-
order behavior in 1-I (kobs = 4.1 × 10−5 s−1). The observed rate
constant (kobs) is inverse first-order in PPh3, implicating PPh3
predissociation from 1-I and reductive elimination from a
short-lived three-coordinate Au(III) complex 3-I under these
conditions (Scheme 3). Consistent with this sequence, PCy3-
supported 2-I did not react at 122 °C over 2 days, presumably
due to the increased donor strength of the trialkylphosphine.
RESULTS AND DISCUSSION
■
Sonicating 1-I or 2-I with excess AgX (X = Br, Cl, F) afforded
metathesis products 1-X and 2-X (X = Br, Cl, F) in high yield
(Scheme 2). Interestingly, 1-F represents a rare example of an
Scheme 2. Synthesis of Au(III) Halide Series
Scheme 3. Proposed General Mechanism of Caryl−X and
Caryl−CF3 Reductive Eliminations from 1-X (X = I, Br, Cl, F)
Figure 1. (A) Thermal ellipsoid representation of 1-F at the 50%
probability level. (B) 19F NMR signal corresponding to the Au−CF3
functionality. (C) 31P{1H} NMR signal in CD2Cl2 corresponding to
the Au-PPh3 functionality.
The zero-order kinetics in the absence of PPh3 suggest
reversible reaction inhibition by starting material. If reductive
elimination proceeds through the coordinatively unsaturated 3-
I, a reasonable origin of this unusual behavior is trapping by 1-I
to μ-iodo bimetallic adduct 4-I (Scheme 3). Indeed, μ-halide
bridges between Au(III) atoms form readily to avoid
coordinative unsaturation at the metal; in addition, bimetallic
complexes such as [AuCl3]2, [Me2AuI]2, [(F3C)2AuX]2 (X = I,
Br), and ([SIPr)Au(Me)F]2)2+ (SIPr = 1,3,-bis(2′,6′-
isolable, terminal organometallic Au(III) fluoride (Figure 1).
All complexes within the 1-X halide series underwent
thermolysis to products of Caryl−X and Caryl−CF3 reductive
elimination, and (when X = F) solvent activation. All reactions
were followed by 19F NMR at 122 °C in toluene-d8 ([1-X] =
14.0−16.0 mM). All values were quantified relative to 1-
Figure 2. (A) Time courses for thermolysis of Au(III)−iodide 1-I with 0, 0.5, 1.0, and 2.0 mol % PPh3. Inset: Time course with 0 mol % added PPh3.
B) Inverse relationship between kobs and [PPh3] indicating inverse first-order behavior of PPh3 in the thermolysis of 1-I.
B
DOI: 10.1021/jacs.5b04613
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 3. (A) Time course for thermolysis of Au(III)−bromide 1-Br in the presence of 9.1−29.9 equiv of Ph3PAuCF3. (B) Direct relationship
between kobs and [Ph3PAuCF3] indicating first-order behavior of Ph3PAuCF3 in the thermolysis of 1-Br.
Figure 4. (A) Time course for thermolysis of Au(III)−chloride 1-Cl in the presence of 8.9−28.4 equiv of Ph3PAuCF3. (B) Direct relationship
between kobs and [Ph3PAuCF3] indicating first-order behavior of Ph3PAuCF3 in the thermolysis of 1-Cl.
diisopropylphenyl)imidazolin-2-ylidene) highlight the steric
and electronic diversity that can complement μ-halide
interactions.10
Treating 3-I as a steady-state intermediate, a complex rate
law consistent with experimental observations can be derived
(eq 1, see Supporting Information for derivation).
elimination were detected after full conversion ([4-Me-C6H4−
Br]/[4-Me-C6H4−CF3] = 1.5:1 for 1-Br, and [4-Me-C6H4−
Cl]/[4-Me-C6H4−CF3] = 1:4.5 for 1-Cl). To our surprise,
reaction rates increased with time for both thermolyses
(Figures 3A and 4A), suggestive of catalysis by products or
nanoparticles.11 Indeed, in the presence of excess Ph3PAuCF3,
the rates of these thermolyses dramatically accelerated,
behaving first-order in 1-Br or 1-Cl and Ph3PAuCF3 (Figures
3B and 4B, and see Supporting Information). The addition of
0.01 equiv (0.014 mM) PPh3 dramatically slowed thermolysis
of 1-Br and 1-Cl with and without excess Ph3PAuCF3,
consistent with phosphine dissociation preceding reductive
elimination in both the nonaccelerated and product-accelerated
pathways.
k
C−I(k1[1‐I] + k−2[4‐I])
d[1‐I]
dt
−
=
k−1[PPh3] + k2[1‐I] + kC−I
(1)
Since [4-I] must be less than [3-I], the assumption that [4-I]
≈ 0 is valid. If the formation of 4-I is significantly faster than
the recombination of PPh3 and 3-I, then k2[1-I] ≫ k−1[PPh3]
+ kC−I, and eq 1 simplifies to the zero-order rate law −d[1-I]/dt
= kC−Ik1/k2, which at 122 °C, is 4.5 × 10−5 M s−1.
Consistent with at least two processes with different product-
determining steps, the ratios [4-Me-C6H4−X]/[4-Me-C6H4−
CF3] vary over time during the thermolyses of 1-Br and 1-Cl.
For instance, when t < 20 min, the accelerated pathway had not
significantly contributed to consumption of 1-Br, and there was
almost no kinetic preference for Caryl−Br or Caryl−CF3 bond
formation ([4-Me-C6H4−Br]/[4-Me-C6H4−CF3] is roughly
1:1). However, in the presence of a large excess of Ph3PAuCF3
(140 mM), the accelerated pathway dominated even at early
reaction times, and Caryl−Br reductive elimination was slightly
favored (2.3:1, presumably the intrinsic kinetic product
distribution of the accelerated pathway.) For 1-Cl, the product
ratio [4-Me-C6H4−Cl]/[4-Me-C6H4−CF3] for the nonaccel-
erated pathway was roughly 1:2.8, while the accelerated
In steady-state, [PPh3] must be very low. Since even small
amounts of PPh3 dramatically alter the reaction behavior, k−1
must be substantially larger than k2. Therefore, when PPh3 is
added, the rate law simplifies to
kC−Ik1[1‐I]
k−1[PPh3]
d[1‐I]
dt
−
=
(2)
where kC−Ik1/k−1 = 2.9 × 10−8 M s−1. Thus, k−1 = (1600)k2, in
accordance with our previous conclusion that k−1 ≫ k2.
Thermolyses of Au(III)−Bromide 1-Br and Au(III)−
Chloride 1-Cl. Qualitatively, the thermolyses of 1-Br and 1-Cl
were notably slower (t1/2 ∼ 75 and 400 min, respectively) than
1-I, and products of both Caryl−X and Caryl−CF3 reductive
C
DOI: 10.1021/jacs.5b04613
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
pathway heavily favored Caryl−CF3 reductive elimination
(1:7.6).
We propose that the electron-withdrawing effect12 of the CF3
ligand renders Ph3PAuCF3 sufficiently Lewis acidic to
coordinate the halide of 1-Br or 1-Cl in a μ-bridging fashion,13
effectively withdrawing electron density from the Au(III) center
and perturbing the relative kinetic preferences for Caryl−X and
Caryl−CF3 reductive elimination from 6-X. Inhibition by PPh3,
the absence of saturation behavior at high [Ph3PAuCF3], and
unobservable intermediates suggest a process involving fast,
reversible coordination of Ph3PAuCF3 to 1-Br or 1-Cl, followed
by PPh3 dissociation and slow Caryl−X and Caryl−CF3 reductive
elimination (Scheme 4).
Figure 5. Time course for thermolysis of Au(III)−fluoride 1-F
a
exhibiting product catalysis. Obtained by monitoring [Ph3PAuCF3].
Scheme 4. Proposed Mechanism for Accelerated
Thermolysis of 1-X (X = Br, Cl)
essentially constant (∼8.3 M at 122 °C in a sealed tube),16
and the ratio of products expressed as rate terms kC−CF3
/
(kAr[toluene-d8]) is also constant (3.6) (Scheme 5). That 3-F
Scheme 5. Proposed Caryl−CF3 and Caryl−Caryl Coupling
Reactions of Thermolysis Intermediate 3-F
can activate solvent implicates an ionic Au(III)−F bond that
imparts sufficient Lewis acidity for formal C−H activation by
electrophilic aromatic substitution, fluoride-assisted deprotona-
tion, or σ-bond metathesis.17
Like 1-I, addition of 0.1 equiv PPh3 (1.4 mM) slowed the
reaction (t1/2 = 300 min) and altered the order in 1-F from zero
to first (see Supporting Information ). However, only biaryl-d7
was formed under these conditions, suggesting an alternative,
slower solvent activation pathway that does not involve 3-F.
Although the Au(III) center in 1-F is less electron-deficient and
more sterically shielded than in 3-F due to coordinative
saturation, it may still be sufficiently Lewis acidic to activate
solvent (Scheme 6). Consistent with this proposal, the reaction
For both 1-Br and 1-Cl, kinetic details of the nonaccelerated
pathway were masked by the accelerated reaction. However, the
slower pathway is likely analogous to 1-I thermolysis (Scheme
3), given the reaction’s sensitivity to excess phosphine and the
diversity of Au(III)-supported μ-halide bridges.11 The un-
ambiguous first-order behavior in the presence of excess
Ph3PAuCF3 clearly indicates that the accelerated reaction is
substantially faster than the nonaccelerated process (see
Supporting Information for rate laws.)
Thermolysis of Au(III)−Fluoride 1-F. The thermolysis of
1-F was slower (t1/2 = 33 min) than that of 1-I, but significantly
faster than that of 1-Br and 1-Cl. Consistent with the apparent
trend of decreasing selectivity of Caryl−X reductive elimination
in the order X = I > Br > Cl, we observed no 4-Me-C6H4−F
upon heating 1-F. Instead, 4-Me-C6H4−CF3 was the major
product. The formation of significant amounts of d7
isotopologues of 2,4′-, 3,4′-, and 4,4′-dimethylbiphenyl
(biaryl-d7) and equimolar Ph3PAuCF3 suggest competitive
activation of toluene-d8 solvent and Caryl−Caryl reductive
elimination from a putative species Au(4-MeC6H4)(aryl-
d7)(CF3).14,15 Since the ratio [4-Me-C6H4−CF3]/[biaryl-d7]
remained constant (3.6:1) throughout the reaction, the rate
laws for both Caryl−CF3 and Caryl−Caryl reductive elimination
must have the same molecularity to first approximation.
Although the selective Caryl−I reductive elimination from 1-I
stands in contrast to the selective Caryl−CF3 reductive
elimination from 1-F, the kinetic behavior for both thermolyses
are notably similar. For instance, the thermolysis of 1-F
exhibited zero-order behavior (up to 80% conversion) (Figure
5) and was dramatically inhibited by PPh3, consistent with slow
Caryl−CF3 reductive elimination and slow solvent activation
from three-coordinate intermediate 3-F, which can be trapped
by starting material (Scheme 3). Although solvent activation is
in all likelihood a bimolecular process, [toluene-d8] is
Scheme 6. Proposed Mechanism of Solvent Activation and
Caryl−Caryl Coupling by 1-F
rate was independent of [PPh3] (from 1.4 to 14 mM), and the
more electron-rich, sterically encumbered 2-F did not react
with toluene-d8.
A rate law consistent with the mechanism of 1-F thermolysis
is shown in eq 3 where the zero-order term is significantly
larger than the pseudo-first-order term in the absence of PPh3,
and k1(kC−CF3 + kAr[tol-d8])/k2 = 3.9 × 10−6 M s−1 (see
Supporting Information for derivation).
k
k2
d[1‐F]
dt
−
=
1 (kC−CF3 + kAr[tol‐d8]) + k′Ar[tol‐d8][1‐F]
(3)
These kinetic investigations reveal that selectivity for Caryl−X
versus Caryl−CF3 reductive elimination from Au(III) decreases
in the order X = I > Br > Cl > F (Figure 6). While rate of Caryl
−
D
DOI: 10.1021/jacs.5b04613
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 6. Distributions of products of reductive elimination from
Au(III) halides 1-X. For 1-Br and 1-Cl, these values represent the
distributions of the nonaccelerated pathway.
X bond formation corresponds to halide polarizibility,7
thermodynamic studies were necessary to determine the role
of ground state effects in the reaction selectivities.
Relative Au(III)−X Bond Dissociation Enthalpies (X = I,
Br, Cl). To gain insight into what extent thermodynamics
govern reductive elimination selectivity, van’t Hoff analyses
between 2-X and trityl halides were carried out. The halide
metathesis equilibria were monitored in toluene-d8 by 19F NMR
at temperatures between 25 and 78 °C. Complexes 2-I and 2-
Br were treated with an excess of Ph3C−Cl (30 equiv) to
ensure fast approach to equilibrium, and to hold [Ph3C−Cl]
constant for determination of the equilibrium constant.18 The
equilibrium between 2-Cl (+ Ph3C−I) and 2-I (+ Ph3C−Cl)
was moderately exothermic (ΔHo = −4.8 kcal/mol) with a
negligible loss of entropy (ΔSo = −2.1 e.u.) (Figure 7).
Figure 8. van’t Hoff plot of the equilibrium of 2-Cl (+ Ph3C−Br) and
2-Br (+ Ph3C−Cl) (shown above) in toluene-d8 between 25 and 78
°C. Initial conditions: 2-Br + Ph3C−Cl (30 equiv).
differences in heats of formation (ΔΔHo ) of 2-Cl, 2-Br, and
f
2-I: ΔHo (2-I) is 13 kcal/mol greater than ΔHo (2-Br), and 21
f
f
kcal/mol greater than ΔHo (2-Cl).
f
The differences in bond dissociation energies (ΔBDE) of
each Au(III)−X bond are functions of ΔΔHo (2-X) and BDEs
f
of the diatomic halogens (see Supporting Information for
derivation).20 Although rough approximations, these values
suggest that the Au(III)−I bond in 2-I is 18 kcal/mol weaker
than the Au(III)−Br bond in 2-Br, and 33 kcal/mol weaker
than the Au(III)−Cl bond in 2-Cl.21 The trend in Au(III)−X
bond strengths follows Caryl−X bond strengths, with the
variation in Au(III)−X BDEs only slightly greater. That the
bond dissociation energies decrease in the order Au(III)−Cl >
Au(III)−Br > Au(III)−I suggests that selectivities for Caryl−X
and Caryl−CF3 reductive elimination are strongly influenced by
the strength of the Au(III)−X bond in the starting material
(Figure 8), and that Au−X bonding must be substantially
diminished in the transition state to Caryl−X reductive
elimination. Halide polarizability, or softness, is correlated
with nucleophilicity, and may also play a role in dictating
relative rates of Caryl−X bond formation, as noted by Hartwig
for Pd(II) systems.2
CONCLUSIONS
■
We have accessed full Au(III) halide families through formal
oxidative addition of CF3I to Au(I) followed by halide
metathesis, and have systematically studied the thermolysis of
1-X (X = F, Cl, Br, I) and the competitive Caryl−X and Caryl
−
Figure 7. van’t Hoff plot of the equilibrium of 2-Cl (+ Ph3C−I) and 2-
I (+ Ph3C−Cl) (shown above) in toluene-d8 between 25 and 78 °C.
Initial conditions: 2-I + Ph3C−Cl (30 equiv).
CF3 reductive eliminations from Au(III). The mechanisms and
kinetic selectivities for these steps are highly dependent on the
identity of the halide ligand. When X = I, thermolysis
exclusively generates the products of Caryl−I bond formation.
The selectivity for Caryl−CF3 reductive elimination increases in
the order X = I < Br < Cl < F, and is completely selective for
Caryl−CF3 bond formation when X = F (Figure 6).
Thermodynamic studies reveal that the Au(III)−X bond
strength increases in the order X = I < Br < Cl, a trend that
mirrors selectivity for Caryl−CF3 reductive elimination. These
Similarly, the equilibrium between 2-Cl (+ Ph3C−Br) and 2-Br
(+ Ph3C−Cl) also lies to the right (ΔHo = −3.1 kcal/mol) with
a negligible entropy loss (ΔSo = −1.8 e.u.) (Figure 8).
Using the thermodynamic parameters above, and differences
in Benson group increments for tertiary alkyl halide groups (see
Supporting Information for derivation),19 we obtain the
E
DOI: 10.1021/jacs.5b04613
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Halide Metathesis between 1-I or 2-I with AgX. 1-I (75 mg,
0.10 mmol) or 2-I (77 mg, 0.10 mmol) was dissolved in CH2Cl2 (5
mL) in a vial. AgX (X = Br, Cl, F) (1.0 mmol) was added at once, and
the reaction was capped and sonicated for 5 min in the dark, followed
by a second addition of AgX (1.0 mmol) and further sonication for 5
min. When X = Br or Cl, the solid turned increasingly yellow with the
formation of AgI. The suspension was filtered through a bed of Celite
and concentrated in vacuo to a white powder that was recrystallized
twice in 1:3 CH2Cl2/pentane to afford 1-Br (52 mg, 0.074 mmol), 2-
Br (61 mg, 0.081 mmol), 1-Cl (51 mg, 0.078 mmol), 2-Cl (60 mg,
0.089 mmol), 1-F (45 mg, 0.071 mmol), or 2-F (55 mg, 0.083 mmol)
in analytical purity as white solids.
observations suggest that selectivity for reductive elimination is
strongly dictated by the Au(III)−X bond strength in the
reactant, and possibly halide polarizability. Highlighting stark
reactivity differences between fluoride and higher halide ligands,
we have also shown that the Au(III)−F bond is relatively ionic,
and can activate C−H/D bonds in arene solvent at elevated
temperatures. Surprisingly, the thermolyses of 1-Br and 1-Cl
are accelerated by Ph3PAuCF3, presumably via coordination of
Ph3PAuCF3 to the Au(III)−bound halide.
In conclusion, Caryl−X reductive elimination can be facile
from Au(III) at elevated temperatures, a process that is rarely
observable and probed systematically at other d8 metal
centers.2,3 Depending on the nature of the halide ligand, this
process can outcompete Caryl−CF3 bond formation. Thus,
irreversible Caryl−X reductive elimination should not be
discounted as a possible, deleterious thermodynamic sink in
studies of organometallic Au(III) halides or Au(I) under
oxidative conditions. These studies also suggest that challenging
Caryl−C reductive elimination from Au(III) halides is favored
when X = Cl or F, due to relatively stronger Au(III)−X bonds
compared to the higher halides. More broadly, reductive
elimination is a fundamental step in many catalytic cycles, and
judicious choice of halide, often considered a spectator ligand,
may in fact be essential to achieving challenging C−C bond
formation.
(Ph3P)Au(4-Me-C6H4)(CF3)(Br) (1-Br). 1H NMR (CD2Cl2, 500
MHz, δ): 7.54−7.49 (m, 3H), 7.46−7.35 (m, 12H), 6.77 (d, J = 8.4
Hz, 2H), 6.64 (d, J = 8.4 Hz, 2H), 2.15 (s, 3H). 13C{1H} NMR
(CD2Cl2, 125 MHz, δ): 135.9, 134.9 (d, J = 10 Hz), 132.2 (d, J = 3
Hz), 130.7 (d, J = 3 Hz), 130.6, 129.1 (d, J = 11 Hz), 126.2, 125.7,
20.6. ipso-13C signals not observed due to heteroatom coupling.
3
31P{1H} NMR (CD2Cl2, 162 MHz, δ): 24.2 (q, JP−F = 68 Hz). 19F
3
NMR (CD2Cl2, 376 MHz, δ): −27.6 (d, JP−F = 68 Hz). Anal. Calcd
for C26H22AuBrF3P: C, 44.66; H, 3.17. Found: C, 44.94; H, 3.33.
1
(Cy3P)Au(4-F-C6H4)(CF3)(Br) (2-Br). H NMR (CD2Cl2, 500 MHz,
δ): 7.31−7.26 (m, 2H), 7.01−6.96 (m, 2H), 2.38−2.26 (m, 3H),
1.91−1.76 (m, 12H), 1.73−1.55 (m, 9H), 1.32−1.20 (m, 3H), 1.14−
1.00 (m, 6H). 13C{1H} NMR (CD2Cl2, 125 MHz, δ): 161.7 (d, J =
246 Hz), 139.3−139.1 (m), 133.0 (dd, J = 6 Hz, J = 1 Hz), 116.3 (d, J
= 20 Hz), 34.1 (d, J = 25 Hz), 29.8 (d, J = 3 Hz), 27.6 (d, J = 11 Hz),
26.3 (d, J = 1 Hz). ipso-13C signals not observed due to heteroatom
3
coupling. 31P{1H} NMR (CD2Cl2, 162 MHz, δ): 28.0 (q, JP−F = 64
Hz). 19F NMR (CD2Cl2, 376 MHz, δ): −29.5 (d, 3JP−F = 63 Hz, Au−
CF3), −117.3 − −117.4 (m, Ar-F). Anal. Calcd for C25H37AuBrF4P: C,
41.62; H, 5.17. Found: C, 41.47; H, 5.33.
METHODS
■
General Considerations. Unless otherwise stated, all manipu-
lations were carried out at ambient temperature (20 °C) under an
atmosphere of purified nitrogen in a Vacuum Atmospheres Corp.
glovebox or with a double manifold vacuum line using standard
Schlenk techniques. All glassware was dried at 150 °C for 12 h prior to
use. Solvents were dried by passage through a column of activated
alumina under nitrogen pressure and degassed by sparging with dry
nitrogen. Toluene-d8 was distilled from sodium ketyl. CF3I was
purchased from Oakwood and connected to a double-manifold
vacuum line fitted with Hg manometers to regulate pressure. AgI,
AgBr, and AgCl were prepared by treating AgNO3 with the respective
NaX (X = halide) salt in water at room temperature, then filtering and
drying. AgF was purchased from Strem and used without further
purification. Ph3C−Cl and Ph3C−Br were purchased from Sigma-
Aldrich and used as received. Ph3C−F was prepared according to
literature procedure.22 (Ph3P)Au(4-Me-C6H4)(CF3)(I) (1-I), (Cy3P)-
Au(4-F-C6H4)(CF3)(I) (2-I), and Ph3PAuCF3 were prepared
according to a recent publication from our lab.1
(Ph3P)Au(4-Me-C6H4)(CF3)(Cl) (1-Cl). 1H NMR (CD2Cl2, 500 MHz,
δ): 7.55−7.50 (m, 3H), 7.44−7.35 (m, 12H), 6.79 (d, J = 8.1 Hz, 2H),
6.64 (d, J = 8.1 Hz, 2H), 2.15 (s, 3H). 13C{1H} NMR (CD2Cl2, 125
MHz, δ): 140.0, 134.8 (d, J = 11 Hz), 132.2 (d, J = 3 Hz), 131.0 (d, J =
3 Hz), 130.7, 129.2 (d, J = 11 Hz), 125.8, 125.3, 20.6. ipso-13C signals
not observed due to heteroatom coupling. 31P{1H} NMR (CD2Cl2,
3
162 MHz, δ): 25.6 (q, JP−F = 69 Hz). 19F NMR (CD2Cl2, 376 MHz,
3
δ): −30.5 (d, JP−F = 69 Hz). Anal. Calcd for C26H22AuClF3P: C,
47.69; H, 3.39. Found: C, 47.75; H, 3.51.
1
(Cy3P)Au(4-F-C6H4)(CF3)(Cl) (2-Cl). H NMR (CD2Cl2, 500 MHz,
δ): 7.34−7.30 (m, 2H), 7.00−6.96 (m, 2H), 2.33−2.22 (m, 3H),
1.90−1.76 (m, 12H), 1.73−1.58 (m, 9H), 1.32−1.21 (m, 3H), 1.14−
1.02 (m, 6H). 13C{1H} NMR (CD2Cl2, 125 MHz, δ): 161.6 (d, J =
243 Hz), 136.6−136.4 (m), 133.2 (dd, J = 7 Hz, J = 1 Hz), 116.4 (d, J
= 21 Hz), 33.5 (d, J = 25 Hz), 29.6 (d, J = 2 Hz), 27.7 (d, J = 11 Hz),
26.3 (d, J = 1 Hz). ipso-13C signals not observed due to heteroatom
3
coupling. 31P{1H} NMR (CD2Cl2, 162 MHz, δ): 28.7 (q, JP−F = 64
NMR spectra were recorded using Bruker AVQ-400, DRX-500, AV-
500 or AV-600 spectrometers, and chemical shifts are referenced to
residual NMR solvent peaks (1H and 13C), 1-CF3-naphthalene (19F),
or H3PO4 (31P). Elemental analyses were performed at the College of
Chemistry Microanalytical Laboratory, University of California,
Berkeley. X-ray structural determinations were performed at CheXray,
University of California, Berkeley on Bruker SMART 1000 or SMART
APEX diffractometers.
Improved Procedure for the Synthesis of 1-I and 2-I. A 25
mL Pyrex Schlenk tube was charged with Ph3PAu(4-Me-C6H4) or
Cy3PAu(4-F-C6H4) (up to 3 mmol) and the solid was dissolved in
CH2Cl2 to give a 0.2 M solution. The tube was sealed and degassed
with three freeze−pump−thaw cycles. CF3I gas was introduced (1
atm) and the reaction vessel was sealed and placed in direct sunlight
for 15 min. The reaction mixture turned yellow within seconds of
irradiation. After irradiation, the excess CF3I was vented and the
reaction mixture adsorbed to neutral alumina and concentrated to
dryness. The alumina mixture was then loaded onto a silica column
and the desired Au(III) compounds were eluted in benzene/hexanes
(1:1 (v/v), Rf = 0.2 for 1-I; Rf = 0.55 for 2-I). Yields typically range
between 60 and 90%. All spectroscopic data match those previously
reported.1
Hz). 19F NMR (CD2Cl2, 376 MHz, δ): −32.8 (d, 3JP−F = 64 Hz, Au−
CF3), −117.4 − −117.5 (m, Ar-F). Anal. Calcd for C25H37AuClF4P: C,
44.36; H, 5.51. Found: C, 44.29; H, 5.40.
1
(Ph3P)Au(4-Me-C6H4)(CF3)(F) (1-F). H NMR (CD2Cl2, 500 MHz,
δ): 7.56−7.51 (m, 3H), 7.49−7.43 (m, 6H), 7.43−7.37 (m, 6H), 6.77
(dd, J = 8.2 Hz, J = 3.3 Hz, 2H), 6.60 (d, J = 8.0 Hz, 2H), 2.15 (s, 3H).
13C{1H} NMR (CD2Cl2, 125 MHz, δ): 135.9, 134.6 (dd, J = 11 Hz, J
= 2 Hz), 132.4 (d, J = 3 Hz), 131.3 (dd, J = 5 Hz, J = 2 Hz), 130.2 (d, J
= 5 Hz), 129.4 (d, J = 11 Hz), 125.5, 125.0, 20.6. ipso-13C signals not
observed due to heteroatom coupling. 31P{1H} NMR (CD2Cl2, 162
MHz, δ): 25.4 (qd, 3JP−F = 69 Hz, 2JP−F = 28 Hz). 19F NMR (CD2Cl2,
3
3
376 MHz, δ): −36.6 (dd, JP−F = 70 Hz, JF−F = 13 Hz), −236.4 −
−236.6 (m). Anal. Calcd for C26H22AuF4P: C, 48.92; H, 3.47. Found:
C, 48.64; H, 3.68.
(Cy3P)Au(4-F-C6H4)(CF3)(F) (2-F). 1H NMR (CD2Cl2, 500 MHz, δ):
7.30−7.24 (m, 2H), 6.96−6.90 (m, 2H), 2.24−2.12 (m, 3H), 1.92−
1.76 (m, 12H), 1.75−1.53 (m, 9H), 1.34−1.21 (m, 3H), 1.17−1.05
(m, 6H). 13C{1H} NMR (CD2Cl2, 125 MHz, δ): 161.7 (d, J = 244
Hz), 133.4−133.2 (m), 116.1 (dd, J = 21 Hz, J = 5 Hz), 32.6 (d, J = 24
Hz), 29.4 (d, J = 2 Hz), 27.6 (d, J = 11 Hz), 26.2 (d, J = 1 Hz) ispo-13C
signals not observed due to heteroatom coupling. 31P{1H} NMR
F
DOI: 10.1021/jacs.5b04613
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
3
2
(CD2Cl2, 162 MHz, δ): 33.2 (qd, JP−F = 65 Hz, JP−F = 23 Hz). 19F
2003, 125, 2890. (c) Casitas, A.; Poater, A.; Sola,
̀
M.; Stahl, S. S.;
3
3
NMR (CD2Cl2, 376 MHz, δ): −39.3 (dd, JP−F = 64 Hz, JF−F = 13
Hz, Au−CF3), −117.8 − −117.9 (m, Ar-F), −249.0 − −249.2 (m).
Anal. Calcd for C25H37AuF5P: C, 45.46; H, 5.65. Found: C, 45.21; H,
5.36.
Costas, M.; Ribas, X. Dalton Trans. 2010, 39, 10458. (d) Lin, B.-L.;
Kang, P.; Stack, T. D. P. Organometallics 2010, 29, 3683. (e) For a
review see: Sheppard, T. D. Org. Biomol. Chem. 2009, 7, 1043.
(4) (a) Dekleva, T. W.; Forster, D. Adv. Catal. 1986, 34, 81.
(b) Goldberg, K. I.; Yan, J.; Winter, E. L. J. Am. Chem. Soc. 1994, 116,
1573. (c) Goldberg, K. I.; Yan, J.; Breitung, E. M. J. Am. Chem. Soc.
1995, 117, 6889. (d) Maitlis, P. M.; Haynes, A.; Sunley, G. J.; Howard,
M. J. J. Chem. Soc., Dalton Trans. 1996, 2187. (e) Frech, C. M.;
Milstein, D. J. Am. Chem. Soc. 2006, 128, 12434. (f) Whitfield, S. R.;
Sanford, M. S. J. Am. Chem. Soc. 2007, 129, 15142. (g) Furuya, T.;
Ritter, T. J. Am. Chem. Soc. 2008, 130, 10060. (h) Racowski, J. M.;
Gary, B. G.; Sanford, M. S. Angew. Chem., Int. Ed. 2012, 51, 3414.
(i) O’Reilly, M. E.; Pahls, D. R.; Cundari, T. R.; Gunnoe, T. B.
Kinetic Experiments. A 14−16 mM solution of 1-X in tol-d8 was
prepared in an inert atmosphere glovebox. Standard (1-trifluorome-
thylnaphthalene) was added by microsyringe, and 500 μL aliquots of
the solution were transferred to oven-dried NMR tubes. The tubes
were capped with greased rubber septa and sealed with Teflon tape.
When appropriate, PPh3 or Ph3PAuCF3 were added directly to the
NMR tube as a solid prior to injection of the tol-d8 solution of 1-X and
standard.
The thermolyses of 1-I and 1-F were carried out in a Bruker DRX-
500 NMR probe that was temperature calibrated using ethylene glycol
and preheated to 122 °C for 30 min. The spectrometer was shimmed
and tuned with a solution of standard, then the NMR tube containing
the solution of interest was lowered into the probe. All other reactions
were carried out at 122 °C in an oil bath shielded from light and the
samples were periodically removed from the bath, cooled to room
temperature, and monitored by 19F NMR.
Thermodynamic Experiments. A 14−16 mM solution 2-X in
tol-d8 was prepared in an inert atmosphere glovebox. Standard (3,5-
ditrifluoromethyl-1-bromobenzene) was added by microsyringe, and
500 μL aliquots of the solution were transferred to oven-dried NMR
tubes charged with Ph3C−Cl (63 mg, 0.23 mmol). The tubes were
capped with greased rubber septa and sealed with Teflon tape. All
experiments were heated in an NMR probe that was calibrated as
described above. The equilibria were first monitored at 25 °C after 10
min at room temperature. After each increase in temperature, the
probe was recalibrated, and the solution of interest was heated in the
probe for 10 min. After equilibrium at maximum temperature (78 °C)
was reached, the reaction was cooled to 25 °C and the equilibrium was
measured.
Organometallics 2014, 33, 6504. (j) Rivada-Wheelaghan, O.; Rosello-
Merino, M.; Díez, J.; Maya, C.; Lopez-Serrano, J.; Conejero, S.
Organometallics 2014, 33, 5944. (k) Perez-Temprano, M. H.;
́
́
́
Racowski, J. M.; Kampf, J. W.; Sanford, M. S. J. Am. Chem. Soc.
2014, 136, 4097. (l) Camasso, N. M.; Sanford, M. S. Science 2015, 347,
1218.
(5) For examples, see: (a) Qiu, D.; Mo, F.; Zheng, Z.; Zhang, Y.;
Wang, J. Org. Lett. 2010, 12, 5475. (b) Nguyen, K. H.; Tomasi, S.; Le
Roch, M.; Toupet, L.; Renault, J.; Uriac, P.; Gouault, N. J. Org. Chem.
2013, 78, 7809.
(6) Mankad, N. P.; Toste, F. D. Chem. Sci. 2012, 3, 72.
(7) Scott, V. J.; Labinger, J. A.; Bercaw, J. E. Organometallics 2010, 29,
4090.
(8) See references contained in: Toste, F. D.; Michelet, V., Eds. Gold
Catalysis: An Homogeneous Approach; Imperial College Press: London,
U.K., 2014.
(9) Relative solvatochromic values (ETN) are 0.099 (toluene) and
0.324 (nitrobenzene). For a review, see: Reichardt, C. Chem. Rev.
1994, 94, 2319.
(10) For examples, see: (a) Margrave, J. L.; Whitmire, K. H.; Hauge,
R. H.; Norem, N. T. Inorg. Chem. 1990, 29, 3252. (b) Zharkova, G. I.;
Baldina, I. A.; Igumenov, I. K. J. Struct. Chem. 2007, 48, 108.
(c) Mankad, N. P.; Toste, F. D. J. Am. Chem. Soc. 2010, 132, 12859.
(11) To discount heterogeneous effects, nanoparticles isolated by
heating 1-Br, 1-Cl or Ph3PAuCF3 at 122 °C over 3 days did not
accelerate the reaction, nor did any solids (unobservable to the naked
eye) centrifuged from a reaction halfway to completion. Filtration of
reaction mixtures (0.2 μm PTFE filter) at 50% conversion also did not
affect the reaction rates when thermolyses were resumed.
(12) The anodic peak potential of (SIPr)CuCF3 is 0.59 V greater
than the potential of (SIPr)CuCH3. Likewise, the anodic peak
potential of (dtbpe)Ni(CF3)2 is 1.17 V greater than the potential of
(dtbpe)Ni(CH3)2. See: Kieltsch, I.; Dubinina, G. G.; Hamacher, C.;
Kaiser, A.; Torres-Nieto, J.; Hutchison, J. M.; Klein, A.; Budnikova, Y.;
Vicic, D. A. Organometallics 2010, 29, 1451.
ASSOCIATED CONTENT
* Supporting Information
Experimental details, characterization data, and crystallographic
information (cif). The Supporting Information is available free
of charge on the ACS Publications website at DOI: 10.1021/
jacs.5b04613.
■
S
AUTHOR INFORMATION
Corresponding Author
*fdtoste@berkeley.edu
Notes
■
The authors declare no competing financial interest.
(13) Three-coordinate Au(I) complexes are well-known, and may be
ACKNOWLEDGMENTS
■
coordinatively fluxional in solution. See: Concepcion
́
Gimeno, M.;
We gratefully acknowledge Professor Robert G. Bergman,
Andrew Samant, and David Kaphan for helpful discussions.
This work was generously funded by the NIHGMS (RO1
GM073932), a NIH fellowship to M.S.W. (F32 GM103238-02)
and an NSF fellowship to W.J.W. (DGE 1106400). X-ray
crystallography was performed using the UC Berkeley College
of Chemistry CheXray facility supported by the NIH Shared
Instrumentation Grant S10- RR027172; we thank Guoqing
Geng and the Chemistry 208 class at UC Berkeley for
assistance with the X-ray structure of 1-Cl.
Laguna, A. Chem. Rev. 1997, 97, 511.
(14) (a) Vicente, J.; Bermudez, M. D.; Escribano, J.; Carrillo, M. P.;
Jones, P. G. J. Chem. Soc., Dalton Trans. 1990, 3083. (b) Vicente, J.;
Bermudez, M. D.; Escribano, J. Organometallics 1991, 10, 3380.
(c) Wolf, W. J.; Winston, M. S.; Toste, F. D. Nat. Chem. 2014, 6, 159.
(d) Levin, M. D.; Toste, F. D. Angew. Chem., Int. Ed. 2014, 53, 6211.
(15) Ph3PAuF and deutero-fluoric (DF) acid are products of Caryl
−
CF3 reductive elimination and tol-d8 activation, respectively. We
observe several silylfluoride species by 19F NMR resulting from facile
Ph3PAuF ionization and quenching of fluoride and DF with the
borosilicate NMR tube.
(16) McLinden, M. O.; Splett, J. D. J. Res. Natl. Inst. Stand. Technol.
2008, 113, 29.
REFERENCES
■
(17) The mechanism of Caryl−H activation is unclear. Mechanisms
involving electrophilic aromatic substitution have been observed with
Au(III), see: (a) Kar, A.; Mangu, N.; Kaiser, H. M.; Tse, M. K. J.
Organomet. Chem. 2009, 694, 524. (b) Ball, L. T.; Lloyd-Jones, G. C.;
Russell, C. A. J. Am. Chem. Soc. 2014, 136, 254. However, although
(1) Winston, M. S.; Wolf, W. J.; Toste, F. D. J. Am. Chem. Soc. 2014,
136, 7777.
(2) Roy, A. H.; Hartwig, J. F. J. Am. Chem. Soc. 2003, 125, 13944.
(3) (a) Klapars, A.; Buchwald, S. L. J. Am. Chem. Soc. 2002, 124,
14844. (b) Zanon, J.; Klapars, A.; Buchwald, S. L. J. Am. Chem. Soc.
G
DOI: 10.1021/jacs.5b04613
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
PhNO2 is substantially less active toward electrophilic aromatic
substitution, selectivity for arene activation in PhNO2 increases by a
factor of 18 (see Supporting Information). Given the greater acidity of
aromatic protons of PhNO2 relative to the deuterons of toluene-d8, we
favor a mechanism involving fluoride-assisted deprotonation. We
observe no benzylic activation of toluene-d8, suggesting that activation
at the arene is kinetically preferred. Such selectivity has been observed
in σ-bond metatheses at late metals, see: (c) Butschke, B.; Schwarz, H.
Organometallics 2011, 30, 1588.
(18) Since [Ph3C−X] = [2-Cl], the equilibrium constant expression
simplifies to Keq = [2-X][Ph3C−Cl]o/[2-Cl]2, and can be solved from
the relative intensities of 19F NMR signals for 2-Cl and 2-Br or 2-I.
(19) Domalski, E. S.; Hearing, E. D. J. Phys. Chem. Ref. Data 1993,
22, 805.
(20) CRC Handbook of Chemistry and Physics, 95th ed.; Haynes, W.
M., Ed.; CRC Press: Boca Raton, FL, 2015; Section 9, pp 65−68.
(21) The qualitative trend of decreasing Au(III)−X BDEs with
higher halides is in marked contrast with gas-phase diatomic Au(I)−X
BDEs, which are lowest for Au(I)−Br and highest for Au(I)−Cl and
Au(I)−I, see: Reuben Brown, J.; Schwerdtfeger, P.; Schroder, D.;
̈
Schwarz, H. J. Am. Soc. Mass Spectrom. 2002, 13, 485.
(22) Oishi, M.; Yamamoto, H. Bull. Chem. Soc. Jpn. 2001, 74, 1445.
H
DOI: 10.1021/jacs.5b04613
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX