Gold(I)-catalysed direct thioetherifications using allylic alcohols: An experimental and computational study

Article abstract of DOI:10.1002/chem.201403293

A gold(I)-catalysed direct thioetherification reaction between allylic alcohols and thiols is presented. The reaction is generally highly regioselective (S_N2′). This dehydrative allylation procedure is very mild and atom economical, producing o

Full text of DOI:10.1002/chem.201403293

DOI: 10.1002/chem.201403293
Full Paper
&
Gold Catalysis
Gold(I)-Catalysed Direct Thioetherifications Using Allylic Alcohols:
an Experimental and Computational Study
Lorena Herkert, Samantha L. J. Green, Graeme Barker, David G. Johnson, Paul C. Young,
Stuart A. Macgregor,* and Ai-Lan Lee*[a]
Abstract: A gold(I)-catalysed direct thioetherification reac-
tion between allylic alcohols and thiols is presented. The re-
action is generally highly regioselective (SN2’). This dehydra-
tive allylation procedure is very mild and atom economical,
producing only water as the by-product and avoiding any
unnecessary waste/steps associated with installing a leaving
or activating group on the substrate. Computational studies
are presented to gain insight into the mechanism of the re-
action. Calculations indicate that the regioselectivity is under
equilibrium control and is ultimately dictated by the thermo-
dynamic stability of the products.
Introduction
to mild reaction conditions which are tolerant of various func-
tional groups. Despite the possible deactivation of gold cata-
lysts by thiols, the potential to develop a mild, selective and
atom economical thioetherification procedure led us to explore
the title reaction.
Thioether linkages are present in many pharmaceuticals, natu-
ral products and synthetic intermediates, and as such, thioeth-
ers are compounds of both academic and industrial impor-
tance.[1] Allylic thioethers also play an important role in bioor-
ganic chemistry[2] and as building blocks for organic synthe-
sis.[3] Despite this, reports using thiols as nucleophiles with late
transition metals are nowhere near as numerous as with other
nucleophiles, because sulfur is known to deactivate/poison
late transition metal catalysts.[4] In the active field of homoge-
nous gold-catalysis,[5] for example, successful reports of thiols
as nucleophiles are relatively scarce.[6,7] Indeed, we have re-
cently shown that thiols in the presence of cationic gold(I) cat-
alysts are in equilibrium with an inactive digold species
[{Au(L)}2(m-SR)]X, which can slow down (or in some cases shut
down) catalytic reactions.[6g,8]
We herein report the first gold-catalysed direct thioetherifi-
cation method using unactivated allylic alcohols and unactivat-
ed thiols as substrates (Scheme 1). This dehydrative allylation
procedure is therefore very atom economical, producing only
water as the by-product and avoiding any unnecessary waste/
steps associated with installing a leaving group on the allyl al-
cohol and activating additive/group on the thiol. Most tradi-
tional metal-catalysed S-allylation methods are not atom effi-
cient, requiring a leaving group and/or thiol activation.[14] More
recently, Ir- and Ru-catalysed dehydrative S-allylation proce-
dures have been developed to overcome this issue, however,
Ir- and Ru-catalysed methods all give the opposite (formal SN2)
regioselectivity, via a presumed p-allyl intermediate.[15] By using
gold(I)-catalysis and thereby accessing a different mechanistic
route (vide infra), direct dehydrative S-allylation by SN2’ regio-
selectivity is now possible for the first time. We also present
a computational study of the mechanism of this process, and
through this account for the regioselectivities observed experi-
mentally: this was found to be dictated by the thermodynamic
stabilities of the products.
One of the research efforts within our group has been to de-
velop gold-catalysed regioselective methods toward allylic
ethers.[9] Within this context, we recently developed a gold(I)-
catalysed direct allylic etherification[10] of unactivated alco-
hols.[11] Unlike previous methods for SN2’ allylic etherifica-
tions,[12] neither the allylic alcohol electrophile[13] nor the alco-
hol nucleophile need to be activated (either to install a leaving
group in the former or form an alkoxide in the latter), leading
[a] L. Herkert, S. L. J. Green, Dr. G. Barker, Dr. D. G. Johnson, P. C. Young,
Prof. Dr. S. A. Macgregor, Dr. A.-L. Lee
Institute of Chemical Sciences, Heriot-Watt University
Edinburgh, EH14 4AS (UK)
E-mail: S.A.Macgregor@hw.ac.uk
A.Lee@hw.ac.uk
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201403293.
Scheme 1. Direct allylic thioetherifications using gold(I)-catalysis.
ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA.
This is an open access article under the terms of the Creative Commons At-
tribution License, which permits use, distribution and reproduction in any
medium, provided the original work is properly cited.
Chem. Eur. J. 2014, 20, 11540 – 11548
11540 ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
acid which may form in the reaction), the reaction still pro-
ceeds well (entry 11), suggesting that the reaction is gold- and
not trace-acid catalysed. Addition of 4 ꢁ molecular sieves was
also investigated (to help remove any water formed during the
reaction) but this appeared to have a detrimental effect on the
conversion (entry 12) and was thus not investigated further.
With our optimised results in hand, a thiol nucleophile
screen was carried out using allylic alcohols 4 as the model
substrate (Table 2). A range of thiophenols 2a–2i reacts
smoothly to give the desired allylic thioether 3a–3i (entries 1–
9). Thiols with substitutuents at the ortho, meta and para posi-
tion perform well (entries 2–6) as do those with electron-do-
nating (entries 2–6 and 9) and electron-withdrawing substitu-
ents (entries 7 and 8). The bromo-substituent in 3h also pro-
vides a handle for further functionalisation. Next, a competing
functional group (OH in 4-mercaptophenol 2i) was investigat-
ed, as phenols have been shown to form chromans or Friedel–
Crafts allylation products under gold-catalysis with allylic alco-
hols.[11a,17] Pleasingly, the reaction proceeds chemoselectively at
the thiophenol site, although the yield is somewhat lower than
with other thiophenols (49%, entry 9). This could be due to
the unprotected OH group slightly interrupting the H-bonded
6-membered ring transition state (vide infra, Scheme 3).
Results and Discussion
Our investigations commenced with the optimisation of the re-
action conditions using allylic alcohol 4 and thiophenol 2a
(Table 1). We have previously ascertained that Echavarren’s cat-
alyst 5[16] is tolerant of deactivation by sulfur nucleophiles,[6g,8]
so 5 was adopted as our preferred catalyst in these studies.
Since a thiol is more nucleophilic than an alcohol, the thiol nu-
cleophile does not need to be in large excess (c.f. 5 equiv alco-
hol nucleophile for the related etherification reaction to avoid
self-reaction of 1),[11] so 1–1.1 equiv of the thiol nucleophile is
sufficient for direct thioetherifications (Table 1). A slightly
higher temperature of 40 versus 308C is clearly useful for
higher conversions (entries 1 and 2) and 24 h appears to be
enough for a reasonable conversion (entries 2–5). Next, chlori-
nated solvents DCE and chloroform pushed the reaction to
higher conversions (90 and 92% respectively, entries 6 and 7)
but crucially also provided cleaner conversions, with no sign of
the inseparable and unidentified side products which ap-
peared in the toluene reactions. A further temperature screen
using chloroform as solvent (entries 7–9) suggests that 358C is
the ideal compromise between mild conditions and good con-
versions. Thus, the conditions in entry 8, Table 1 were adopted
as the optimised general conditions.
Next, we moved from thiophenols to alkyl thiols (entries 10–
16). This change of S-nucleophile class initially caused several
problems. Firstly, the reactions are a lot slower, despite alkyl
thiols being more nucleophilic. We have recently ascribed this
to the fact that the higher Lewis basicity of alkyl thiols (vs.
thiophenols) will push the equilibrium towards the deactivated
[{Au(L)}2(m-SR)]SbF6 species 6, resulting in a lower concentra-
tion of active catalyst in solution (Scheme 2).[8] Nevertheless,
higher temperatures (50–608C) and longer reaction times
(72 h) allow for successful conversions to the desired allylic
thioether products.
Table 1. Selected optimisation and control reactions.
Having overcome the first reactivity issue, a second problem
soon emerged, especially when primary alkyl thiols were em-
ployed as nucleophiles. Although the desired allylic thioether
(e.g., 3j) is observed in the crude mixture, the product oxidises
upon silica gel chromatography to give the sulfone 7j
(Figure 1) instead (entry 10). Attempts to stop the oxidation by
using an alumina column met with only moderate success
(16% 3j). Since control reactions show that leaving 3j in air,
silica gel or alumina, respectively, does not cause any oxida-
tion, we surmised that it must be a combination of the gold
catalyst, silica and air during column chromatography which
causes the oxidation. Indeed, removing the gold using a resin-
bound scavenger Reaxa QuadraPureTM MPA[10f] prior to purifica-
tion by column chromatography solves the oxidation problem
and allows the allylic thioether 3j to be obtained successfully
in 56% yield (entry 11).
Entry
T [8C]
Solvent[a]
t [h]
Conv. [%][b]
1
2
3
4
5
6
7
8
30
40
40
40
40
40
40
35
30
40
40
40
toluene
toluene
toluene
toluene
toluene
DCE
CHCl3
CHCl3
CHCl3
72
72
8
46[c]
88[c]
67[c]
81[c]
82[c]
90
92
91
79
0
24
48
24
24
24
24
24
24
24
9
10[d]
11[e]
12[f]
toluene
toluene
toluene
76
9
[a] 0.4m. [b] Determined by 1H NMR analysis of the crude mixture. [c] Un-
identified side-products were observed in toluene. [d] No gold catalyst
added. [e] 2,6-Di-tert-butylpyridine (5 mol%) added. [f] 4 ꢁ molecular
sieves added.
The secondary alkyl thiol 2l was less prone to oxidation
(entry 13), implying that the oxidation is sensitive to steric hin-
drance. Nevertheless, the yield of 3l is still improved upon
scavenging the gold catalyst prior to purification by column
chromatography (56 vs. 40%, entries 13 and 14). Finally, che-
moselectivity was probed by using alkyl thiols with pendant
functional groups (entries 15 and 16). The hydroxyl group in
A few control reactions were also carried out in order to as-
certain if gold is acting as a catalyst. In the absence of a gold
catalyst, the reaction does not proceed at all (entry 10). When
a sterically hindered mild base 2,6-di-tert-butylpyridine was
added to the gold-catalysed reaction (to mop up any trace
Chem. Eur. J. 2014, 20, 11540 – 11548
www.chemeurj.org
11541 ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
2n is tolerated, although it produ-
ces a modest 45% yield (entry 15).
As with entry 9, a plausible cause
for this lower yield with 2n is the
disruption of the H bonding in the
6-membered transition state by
the ꢀOH pendant group (vide
Table 2. Thiol nucleophile scope.
Figure 1. Oxidised product 7j.
infra). In contrast, a carboxylic acid pendant group is tolerated
very well (2o, entry 16), providing 3o in 92% yield. It should
be noted, however, that while entries 1–15 all provide exclu-
sively SN2’ regioselectivity, 3o is the only thiol investigated to
show traces of the formal SN2 isomer (~5%). Nevertheless, the
selectivity is still very good (18:1 SN2’/SN2).
Entry
T
Thiol
Product
Yield
[%][a]
[8C]
2
1
2
35
35
3a
3b
70
62
Having established the sulfur nucleophile scope, we turned
our attention to the allylic alcohol substrate scope (Table 3). In-
itially, substituent effects on tertiary allylic alcohols were inves-
tigated (entries 1–9). Increasing the steric bulk of the substitu-
ents (n-hexyl!cyclohexyl) is tolerated and in fact produces
a good 86% yield (entries 1 and 2). Where E/Z isomers are pos-
sible, the reaction is selective for the E isomer (entries 3–5),
with the E/Z ratios increasing the better the steric differentia-
tion between the two substituents (e.g., >20:1 for 3q but
a lower 9:1 for 3r). An exocyclic tertiary allylic alcohol sub-
strate 12 is also tolerated well (entry 6). Next, the effect of sub-
stitution around the alkene was investigated, thus, g-substitut-
ed (entries 7 and 8) and b-substituted (entry 9) tertiary allylic
alcohols were subjected to the reaction conditions. Pleasingly,
both the g-substituted and the b-substituted substrate reacted
to give decent yields (50, 72 and 61%, respectively), with 3v
also providing >20:1 E/Z selectivity.
3
4
35
35
3c
76
71
3d
5
35
3e
66
6
7
8
35
35
35
3 f
64
68
73
3g
3h
Next, secondary allylic alcohols were investigated as sub-
strates (entries 10–15). The secondary allylic alcohols require
a higher temperature (458C) and longer reaction time (72 h) to
go to completion. Nevertheless, cyclic allylic alcohol 16 pro-
vides a very good 89% yield of desired product 3x (entry 10).
Acyclic allylic alcohol 17 as well as g-substituted secondary al-
cohol 18 (both with a Ph substituent) also react smoothly (en-
tries 11 and 12). Replacing the Ph substituent in 18 with alkyl
substituents (19–21), however, provided some of the poorest
selectivities of all the allylic alcohols screened (entries 13–15).
Firstly, the nBu substituted 19 provided a reasonable 8:1
formal SN2’/SN2 regioselectivity, but a very poor 3:2 E/Z selec-
tivity (entry 13). Increasing the steric hindrance from nBu to Cy
(20) gives a good 85% yield and drastically improves the E/Z
selectivity (>20:1) but the regioselectivity drops (SN2’/SN2 2:1,
entry 14). Finally, increasing the bulkiness of the substituent
even further from Cy to tBu (21) provides excellent (>20:1) re-
gioselectivity again, with a decent 4:1 E/Z ratio (entry 15).
Reactions with primary allylic alcohols were significantly less
straightforward (entries 16–18). Both the trans and cis primary
allylic alcohols 22 and 23 were reluctant to undergo the reac-
tion, providing mainly unreacted starting material and a com-
plex mixture of other unidentified products (entries 16 and 17).
In contrast, g,g-disubstituted primary allylic alcohol 24 reacts
smoothly, but gives the opposite formal SN2 regioselectivity
(25 instead of 3, entry 18). In order to ascertain if it is the g,g-
substitution which causes a switch of regioselectivity, or if it is
9
35
35
3i
7j
49
42
10
11[b]
12[b]
50
50
BnSH 2j
nBuSH 2k
3j
3k
56
47
13
60
60
50
3l
40
56
45
14[b]
15[b]
CySH 2l
3l
3n
16[b]
45
3o
92[c]
[a] Isolated yield. Regioselectivity SN2’/SN2>20:1 unless otherwise stated.
[b] Resin-bound scavenger Reaxa QuadraPureTM MPA added at the end of
reaction. [c] Formal SN2’/SN2 ratio 18:1.
Scheme 2. Postulated mechanism for formation of 6 from 5.
Chem. Eur. J. 2014, 20, 11540 – 11548
www.chemeurj.org
11542 ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Table 3. Allylic alcohol scope.
Entry
1
Allylic alcohol
Product
Yield [%][a]
68
Note[b]
–
2
3
4
5
6
7
8
9
86
59
59
73
55
50
72
61
–
>20:1 E/Z
9:1 E/Z
~6:1 E/Z
–
–
>20:1 E/Z
–
10[c]
11[c]
89
64
–
>20:1 E/Z
12[c]
13[c]
14[c]
15[c]
16
72
>20:1 E/Z
65
~3:2 E/Z, 8:1 SN2’/SN2
>20:1 E/Z, 2:1 SN2’/SN2
4:1 E/Z
85
71
–
–
N.D.[d]
N.D.[d]
59
poor conv.
17
poor conv.
18
“SN2” regioselectivity
Chem. Eur. J. 2014, 20, 11540 – 11548
www.chemeurj.org
11543 ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Table 3. (Continued)
Entry
19
Allylic alcohol
Product
Yield [%][a]
87
Note[b]
“SN2” regioselectivity
[a] Isolated yield. Regioselectivity SN2’/SN2>20:1 unless otherwise stated. [b] Determined by 1H NMR analysis. [c] 458C for 72 h. [d] Not determined. Poor
conversions and complex mixture of products.
specific to the primary allylic alcohol 24, the secondary g,g-
substituted alcohol 26 was investigated next (entry 19).
Indeed, the formal SN2 product 27 is formed exclusively in
good 87% yield. Therefore, it appears that although the direct
thioetherification is in general SN2’ selective (Table 2 and en-
tries 1–15, Table 3), g,g-substitution causes a complete switch
in regioselectivity (entries 18 and 19, Table 3).
Having completed our studies on the substrate scope, we
were keen to extend this methodology to asymmetric meth-
ods. Although the majority of the products formed by this
Scheme 3. Plausible mechanisms. Calculations and control studies point
method are achiral, some of the g-substituted substrates may
be amenable to chirality transfer. In theory, an enantioenriched
chiral allylic alcohol with a g-substituent (such as 18, entry 12),
which is easily accessible in high ee values by Sharpless kinetic
resolution[18] or enzyme resolution,[19] should be able to transfer
its chirality[20] to the allylic thioether product 3z, especially if
a 6-membered ring H-bonded intermediate is involved. Disap-
pointingly, our initial attempt at chirality transfer with enan-
tioenriched secondary allylic alcohol (S)-18 (91:1 e.r.)[21] pro-
duced only racemic product 3z. Suspecting that the phenyl
substituent in 18 is to blame (the phenyl substituent may be
stabilising a planar allylic cation intermediate which may be
otherwise disfavoured), enantioenriched allylic alcohol (R)-21
(94:6 e.r.)[19] and (R)-20 (>99:1 e.r.) were investigated instead.
Unfortunately, the corresponding products 3ac and 3ab were
also racemic. These results are surprising as the allylic etherifi-
cation counterpart (alcohol additions to allylic alcohols) is
known to proceed by chirality transfer.[11b,22] Therefore, al-
though the lack of chirality transfer in the above g-substituted
cases is disappointing, it does provide valuable indication that
the factors controlling allylic etherification and allylic thioether-
ification may be subtly different.
toward 1 and 3 being the likely pathways.
A possible alternative mechanism to explain the racemisa-
tion is that the reaction proceeds via an allyl cation D (Eq. (2),
Scheme 3). Gold(I) catalysts are also known to coordinate to al-
cohols,[23] in this case turning the hydroxyl group into a better
leaving group (C). Attack at the less hindered position could
occur on D to provide the product 3. Active catalyst LAu+ is
presumably regenerated by protonolysis of LAuOH.[24] Such
a mechanism could explain why the reaction is easiest with
tertiary allylic alcohols (e.g., 13, 14), slightly harder with secon-
dary (18–21) and does not work with primary (22 and 23; at
least when comparing mono-g-substituted substrates), due to
differences in the stabilisation of the cationic intermediate.
This alternative mechanism, however, is not totally consis-
tent with all observations. For example, if the reaction indeed
goes via an allylic cation D, the reaction with 19 might be ex-
pected to have poor regioselectivity (primary alkyl substituent
at both ends of D), but a good 8:1 formal SN2’/SN2 selectivity is
observed. Therefore, a third alternative might be that the reac-
tion goes through Equation (1), but the product 3 then race-
mises (by isomerisation between formal SN2’ and formal SN2
type products, Eq. (3), Scheme 3), especially in cases where
higher reaction temperatures and times are required to push
the reaction to completion (e.g., entries 10–15, Table 3).
Indeed, the mechanism could also change depending on the
substituents present.
If the direct allylic thioetherification proceeds in a similar
fashion to the etherification procedure,[11a,b] then chirality trans-
fer would be expected (Eq. (1), Scheme 3). Since gold(I) is an
excellent p-Lewis acid,[5e] it is likely to activate the alkene func-
tionality in the allylic alcohol towards attack by a thiol nucleo-
phile (A, Scheme 3).[10g] Demetallation and elimination of water
(possibly enabled by intramolecular H bonding, B), will then re-
generate the catalyst and produce the desired allylic ether
product 3. Such a chair-like 6-membered ring transition state
would be crucial for any chirality transfer (Eq. (1), Scheme 3).
Density functional theory (DFT) calculations have been per-
formed to probe these selectivity issues.[25] The calculations
mainly employed a [(Me3P)Au(MeCN)]+ model catalyst, 5’, and
initially considered the reaction of PhSH with Me2C(OH)CH=
Chem. Eur. J. 2014, 20, 11540 – 11548
www.chemeurj.org
11544 ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
CH2, 4’, a simple model of allylic alcohol 4 with the n-hexyl
groups replaced by methyls. Subsequent calculations with the
full Echavarren catalyst, 5, provided very similar energetics, de-
tails of which are given in the Supporting Information. We
report free energies derived from gas-phase optimisations with
the BP86 functional, corrected for chloroform solvent (PCM
method) and dispersion effects (i.e., at the BP86-D3(CHCl3)
level).
As a first step, we considered the thermodynamics of the al-
ternative formal SN2’ and SN2 products that can be formed
(along with water) through the reaction of 4’ with PhSH
(Scheme 4). Both processes are exergonic, but the formal SN2’
product, 3a’, is computed to be 4 kcalmolꢀ1 more stable than
the alternative formal SN2 product 28.
Figure 2. Computed free energy reaction profile (BP86-D3(CHCl3), kcalmolꢀ1
for the reaction of 4’ with PhSH at model catalyst 5’ to form 3a’ and 28.
)
compared with 2.30 ꢁ in IntI4’. At this stage there is relatively
little elongation of either the S1ꢀH1 (1.39 ꢁ compared with
1.37 ꢁ in IntI4’) or the C3ꢀO bonds (1.47 ꢁ compared with
1.46 ꢁ in IntI4’) and in these terms the transition state exhibits
an early geometry. In contrast TS(III-IV)4’, equating to nucleo-
philic attack of a second PhSH at C3, has a late geometry with
the new C3ꢀS2 bond nearly fully formed (1.94 ꢁ compared with
1.92 ꢁ in IntIV4’), extension of the C1LS1 distance to 2.37 ꢁ and
slippage of the gold away from the C2ꢀC3 bond such that it
begins to interact with C1 (AuLC1 =2.67 ꢁ).
Scheme 4. Thermodynamics of the alternative products derived from 4’:
formal SN2’ product 3a’ and formal SN2 product 28.
A computed mechanism for the formation of 3a’ and 28 is
summarised in Figure 2, where free energies are quoted rela-
tive to the model catalyst 5’ and the separated reactants (4’
and PhSH). The profile starts from IntI4’, formed from 5’ by dis-
placement of MeCN by a p-bound allylic alcohol which is itself
H bonded to the external PhSH. The latter is located anti to
the metal centre and so is set up for external attack of the
thiol at the g-position (C1, see Figure 3 for atom numbering).
This SN2’ step proceeds with concerted proton transfer and
loss of H2O to give IntII4’, where the allylic thioether product is
now p-bound to gold and H2O is H bonded to the sulfur
centre.[26] The process occurs via TS(I-II)4’ with a modest barrier
of only 6.8 kcalmolꢀ1. Subsequent conformational searching re-
vealed several additional forms of IntI4’, the most stable of
which has G=ꢀ2.5 kcalmolꢀ1. This species, IntI4’b, is related to
IntI4’ by rotation about the C2ꢀC3 bond which then places the
{HOLHSPh} moiety over the Au centre. The overall barrier for
the SN2’ process, relative to this low energy precursor, is there-
fore 11.6 kcalmolꢀ1 and so remains readily accessible (assum-
ing facile interconversion of IntI4’b and IntI4’).[27,28] From IntII4’
the SN2’ product 3a’ can then either be released, or alterna-
tively a second SN2’ step with PhSH can lead to the net formal
SN2 product 28. Thus, displacement of H2O from IntII4’ with
PhSH gives IntIII4’ (G=ꢀ3.2 kcalmolꢀ1). Thiol attack via
TS(III-IV)4’ (G= +8.0 kcalmolꢀ1) then generates IntIV4’ (G=
ꢀ1.8 kcalmolꢀ1) in which the net SN2 product 28 is now
p-bound to the gold centre.
Figure 3. Computed geometries of SN2’ transition states TS(I-II)4’ and
TS(III-IV)4’ with selected distances [ꢁ] and free energies [kcalmolꢀ1].
The low barriers and moderate exergonicities associated
with the two SN2’ steps in Figure 2 suggest that both processes
should be readily accessible and reversible at the temperatures
used experimentally. The similar energies of the different p-
bound intermediates also suggest that a mixture of such spe-
cies will be formed in solution during the catalysis, although
IntII4’, the immediate precursor to the observed SN2’ thioether
product (here modeled by 3a’), is the most stable of these.
This is also consistent with the reaction conditions, which re-
The computed structures of TS(I-II)4’ and TS(III-IV)4’ are
given in Figure 3. For TS(I-II)4’ the main geometric feature is
a slipping of the alkene moiety induced by the approach of
the thiol, with an elongation of the AuꢀC1 distance to 2.61 ꢁ
Chem. Eur. J. 2014, 20, 11540 – 11548
www.chemeurj.org
11545 ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
quire relatively long reaction times to achieve equilibrium, but
without significant heating. The experimentally observed selec-
tivity for the SN2’ product reflects therefore the greater ther-
modynamic stability of this species over the alternative formal
SN2 product 28, rather than any kinetic preference for the SN2’
product.
This thermodynamic basis for regioselectivity appears quite
general, for example a clear preference was also computed for
the SN2’ thioetherification products formed between PhSH and
substrates 9, 10 and 11 (models of 3q, 3r and 3s, respectively,
Table 3) over their formal SN2 alternatives. Experimentally, g,g-
disubstituted substrates 24 and 26 were exceptions in forming
SN2 products. Calculations showed that this is consistent with
a change in the relative energies of the products with 3a’
(here a model of 25, the formal SN2 product derived from 24),
being 4.0 kcalmolꢀ1 more stable than the alternative SN2’ prod-
uct 28 (Scheme 5).
Figure 4. Computed free energy reaction profile (BP86-D3(CHCl3)) for the re-
action of 24 with PhSH at model catalyst 5’ to form 28 and 3a’.
Scheme 5. Thermodynamics of the alternative products derived from 24:
formal SN2’ product 28 and formal SN2 product 3a’.
Due to the different regioselectivity in this case we also con-
firmed that the formal SN2 product, 3a’, is readily formed via
two sequential SN2’ steps (Figure 4). Thus, nucleophilic attack
of PhSH with IntI24 proceeds through TS(I-II)24 at
+9.9 kcalmolꢀ1 to give IntII24 in which 28 is bound to the Au
centre. A second SN2’ step with PhSH then occurs via
TS(III-IV)24 (G= +6.4 kcalmolꢀ1) to give IntIV24 at
ꢀ4.8 kcalmolꢀ1 from which 3a’ can be released. The overall
shape of the free energy surface in Figure 4 reiterates the
facile reversibility of these processes, with the immediate pre-
cursor to the observed product (IntIV24) again corresponding
to the most stable intermediate. In this case we also consid-
ered whether the formal SN2 product 3a’ could be formed by
a direct SN2 reaction of PhSH at an O-bound precursor, IntV24
(Figure 5a). IntV24 has a relative free energy of +0.5 kcalmolꢀ1
suggesting it is accessible in solution. However, a computed
barrier via TS(V-VI)24 of 26.3 kcalmolꢀ1 indicates this process is
disfavoured kinetically relative to the facile sequential SN2’
steps shown in Figure 4. Inclusion of an additional PhSH mole-
cule in the calculation (to act as an initial proton acceptor in
the transition state) did not result in any reduction in the barri-
er (see the Supporting Information). Similarly the formation of
an allylic cation (C!D Scheme 3 and Figure 5b) can be dis-
counted as this process is associated with a free energy
change of +28.5 kcalmolꢀ1. We also tested the formation of
the analogous allylic cation from substrate 17 and found that,
while this is somewhat more accessible, it does remain strong-
ly endergonic, by +23.9 kcalmolꢀ1. Allylic cation formation
Figure 5. Alternative pathways for the formation of 3a’ from O-bound pre-
cursor IntVI24 via: a) direct SN2 attack, and b) an allylic cation. Intermediate
free energies at the BP86-D3(CHCl3) level are in kcalmolꢀ1
.
therefore cannot account for the lack of chirality transfer ob-
served with enantioenriched substrate 17, despite the pres-
ence of the potentially stabilising phenyl substituent.
The computational studies point towards accessible isomeri-
sations between the two possible allylic thioether products
(formal SN2’ or SN2 regioselectivities) and that any regioselectiv-
ity is ultimately dictated by the thermodynamic stability of the
products. With this in mind, we postulated that the apparent
failure of enantioenriched (S)-18, (R)-21 and (R)-20 to transfer
chirality to its respective products is due to the subsequent
racemisation of any enantioenriched product that might be
formed (via path 3 in Scheme 3). To gain experimental proof
for these hypotheses, a racemisation/isomerisation experiment
was carried out (Scheme 6). A mixture of 3ab (which would be
the formal SN2’ product in Entry 14, Table 3) and 29 (which
would be the formal SN2 product in entry 14, Table 3) was pre-
Chem. Eur. J. 2014, 20, 11540 – 11548
www.chemeurj.org
11546 ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
sure and the crude material was purified by flash column chroma-
tography. Full experimental procedures, characterisation for all new
1
compounds and copies of H and 13C NMR spectra are provided in
the Supporting Information.
Acknowledgements
Scheme 6. Racemisation and isomerisation study of products under the
gold(I)-catalysed thioetherification reaction conditions.
We thank EPSRC (EP/K00736X/1), EPSRC DTA studentship
(P.C.Y.) and Erasmus Programme (L.H.) for funding. We thank
Nina A. Schopf for preliminary experiments. Mass spectrometry
data were acquired at the EPSRC UK National Mass Spectrome-
try Facility at Swansea University.
pared using an alternative three-step procedure by Gais;[29] this
produced an inseparable mixture of (R)-3ab in 80:20 e.r. and
rac-29 in a 2:1 ratio. This 2:1 mixture was then subjected to
the gold(I) reaction conditions (Scheme 6). Indeed, not only is
the enantioenriched (R)-3ab completely racemised but the
ratio of 3ab/29 also increases from 2:1 to 4:1, suggesting equi-
libration towards the more stable isomer.[30] Therefore, path 1
followed by path 3 is likely the reason for the observed racem-
ised products, rather than path 2 (Scheme 3).
Keywords: allylation · allylic alcohols · gold · mechanism ·
thiols
[1] For example, see: a) J. Clayden, P. MacLellan, Beilstein J. Org. Chem.
2011, 7, 582–595; b) F. Robertson, J. Wu, Org. Lett. 2010, 12, 2668–
2671, and references therein.
[2] For example, see: a) L. Brunsveld, J. Kuhlmann, K. Alexandrov, A. Wit-
tinghofer, R. S. Goody, H. Waldmann, Angew. Chem. 2006, 118, 6774–
6798; Angew. Chem. Int. Ed. 2006, 45, 6622–6646; b) Y. A. Lin, J. M.
Chalker, N. Floyd, G. J. L. Bernardes, B. G. Davis, J. Am. Chem. Soc. 2008,
130, 9642–9643.
[3] Selected reviews: a) A. R. Katritzky, M. Piffl, H. Y. Lang, E. Anders, Chem.
Rev. 1999, 99, 665–722; b) M. D. McReynolds, J. M. Dougherty, P. R.
Hanson, Chem. Rev. 2004, 104, 2239–2258.
Conclusion
In conclusion, a mild, regioselective (formal SN2’) and stereose-
lective (E) gold(I)-catalysed direct thioetherification of allylic al-
cohols has been developed. The reaction has a wide substrate
scope including tertiary, secondary and certain primary allylic
alcohols, and tolerates a wide range of substituents on the
thiol nucleophile. Unlike the related etherification reaction em-
ploying alcohol nucleophiles, the thiol nucleophile does not
need to be in excess; an approximately 1:1 ratio of substrates
is sufficient with no observed self-reaction of the allylic alcohol
substrate. Furthermore, the SN2’ regioselectivity is generally ex-
cellent. The exception to this is when the g-position of the al-
lylic alcohol is fully substituted. When alkyl thiols are used as
nucleophiles, the resulting products are prone to oxidation to
the sulfone during silica gel chromatography, but this is suc-
cessfully avoided by employing a gold-scavenging resin Reaxa
QuadraPureTM MPA prior to purification. Computational studies
on selected substrates suggest that both the formal SN2’ and
SN2 products can be readily and reversibly formed under the
reaction conditions. The observed regioselectivities are there-
fore dictated by the thermodynamic stabilities of the products.
The observation of any formal SN2 products will be the result
of two sequential SN2’ steps, this route being kinetically fav-
oured over a direct SN2 reaction. The lack of chirality transfer in
enantioenriched substrates can also be accounted for by the
racemisation of products.
[4] T. Kondo, T. Mitsudo, Chem. Rev. 2000, 100, 3205–3220.
[5] Selected reviews on gold catalysis: a) M. Bandini, Chem. Soc. Rev. 2011,
40, 1358–1367; b) N. Bongers, N. Krause, Angew. Chem. 2008, 120,
2208–2211; Angew. Chem. Int. Ed. 2008, 47, 2178–2181; c) T. C. Boor-
man, I. Larrosa, Chem. Soc. Rev. 2011, 40, 1910–1925; d) A. Corma, A.
Leyva-Perez, M. J. Sabater, Chem. Rev. 2011, 111, 1657–1712; e) A. Fꢂrst-
ner, P. W. Davies, Angew. Chem. 2007, 119, 3478–3519; Angew. Chem.
Int. Ed. 2007, 46, 3410–3449; f) D. J. Gorin, B. D. Sherry, F. D. Toste,
Chem. Rev. 2008, 108, 3351–3378; g) D. J. Gorin, F. D. Toste, Nature
2007, 446, 395–403; h) A. S. K. Hashmi, Chem. Rev. 2007, 107, 3180–
3211; i) A. S. K. Hashmi, M. Buehrle, Aldrichimica Acta 2010, 43, 27–33;
j) A. S. K. Hashmi, G. J. Hutchings, Angew. Chem. 2006, 118, 8064–8105;
Angew. Chem. Int. Ed. 2006, 45, 7896–7936; k) E. Jimꢃnez-NfflÇez, A. M.
Echavarren, Chem. Commun. 2007, 333–346; l) E. Jimꢃnez-NfflÇez, A. M.
Echavarren, Chem. Rev. 2008, 108, 3326–3350; m) Z. G. Li, C. Brouwer, C.
He, Chem. Rev. 2008, 108, 3239–3265; n) N. Marion, S. P. Nolan, Chem.
Soc. Rev. 2008, 37, 1776–1782; o) J. Muzart, Tetrahedron 2008, 64,
5815–5849; p) M. Rudolph, A. S. K. Hashmi, Chem. Soc. Rev. 2012, 41,
2448–2462; q) S. Sengupta, X. Shi, ChemCatChem 2010, 2, 609–619;
r) N. D. Shapiro, F. D. Toste, Synlett 2010, 675–691; s) H. C. Shen, Tetrahe-
dron 2008, 64, 3885–3903; t) H. C. Shen, Tetrahedron 2008, 64, 7847–
7870.
[6] Examples using thiols as nucleophiles: a) K. Ando, J. Org. Chem. 2010,
75, 8516–8521; b) A. Arcadi, G. Bianchi, S. Di Giuseppe, F. Marinelli,
Green Chem. 2003, 5, 64–67; c) C. Brouwer, R. Rahatrian, C. He, Synlett
2007, 1785–1789; d) A. Corma, C. Gonzalez-Arellano, M. Iglesias, F. San-
chez, Appl. Catal. A 2010, 375, 49–54; e) Menggenbateer, M. Narsireddy,
G. Ferrara, N. Nishina, T. Jin, Y. Yamamoto, Tetrahedron Lett. 2010, 51,
4627–4629; f) N. Morita, N. Krause, Angew. Chem. 2006, 118, 1930–
1933; Angew. Chem. Int. Ed. 2006, 45, 1897–1899; g) R. J. Mudd, P. C.
Young, J. A. Jordan-Hore, G. M. Rosair, A.-L. Lee, J. Org. Chem. 2012, 77,
7633–7639; h) S. Biswas, J. S. M. Samec, Chem. Commun. 2012, 48,
6586–6588; i) S. Biswas, R. A. Watile, J. S. Samec, Chem. Eur. J. 2014, 20,
2159–2163.
Experimental Section
General procedure
Thiol 2 (1.1 equiv) and catalyst 5 (5 mol%) were added to a solution
of allylic alcohol (1 equiv) in chloroform (0.4m). The resulting mix-
ture was allowed to stir at 35–608C for 24–72 h as required. The
mixture was then filtered through a plug of silica using diethyl
ether as eluent. The filtrate was concentrated under reduced pres-
[7] Examples of other low-valent sulfur (thioethers) employed as nucleo-
philes in gold-catalysis: a) P. W. Davies, S. J. C. Albrecht, Chem. Commun.
2008, 238–240; b) P. W. Davies, S. J. C. Albrecht, Synlett 2012, 23, 70–
73; c) I. Nakamura, T. Sato, Y. Yamamoto, Angew. Chem. 2006, 118,
Chem. Eur. J. 2014, 20, 11540 – 11548
www.chemeurj.org
11547 ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
4585–4587; Angew. Chem. Int. Ed. 2006, 45, 4473–4475; d) L. Peng, X.
Zhang, S. Zhang, J. Wang, J. Org. Chem. 2007, 72, 1192–1197.
[8] P. C. Young, S. L. J. Green, G. M. Rosair, A.-L. Lee, Dalton Trans. 2013, 42,
9645–9653.
[9] a) J. T. Bauer, M. S. Hadfield, A.-L. Lee, Chem. Commun. 2008, 6405–
6407; b) M. S. Hadfield, J. T. Bauer, P. E. Glen, A.-L. Lee, Org. Biomol.
Chem. 2010, 8, 4090–4095; c) M. S. Hadfield, L. J. L. Haeller, A.-L. Lee,
S. A. Macgregor, J. A. T. O’Neill, A. M. Watson, Org. Biomol. Chem. 2012,
10, 4433–4440; d) M. S. Hadfield, A.-L. Lee, Org. Lett. 2010, 12, 484–
487; e) A. Heuer-Jungemann, R. McLaren, M. Hadfield, A.-L. Lee, Tetrahe-
dron 2011, 67, 1609–1616; f) K. J. Kilpin, U. S. D. Paul, A.-L. Lee, J. D.
Crowley, Chem. Commun. 2011, 47, 328–330.
Pradhan, Y. Maegawa, M. Kitamura, Chem. Commun. 2010, 46, 3996–
3998; d) A. B. Zaitsev, H. F. Caldwell, P. S. Pregosin, L. F. Veiros, Chem. Eur.
J. 2009, 15, 6468–6477.
[16] C. Nieto-Oberhuber, S. Lopez, M. P. MuÇoz, D. J. Cµrdenas, E. Bunuel, C.
Nevado, A. M. Echavarren, Angew. Chem. 2005, 117, 6302–6304; Angew.
Chem. Int. Ed. 2005, 44, 6146–6148.
[17] E. Coutant, P. C. Young, G. Barker, A.-L. Lee, Beilstein J. Org. Chem. 2013,
9, 1797–1806.
[18] V. S. Martin, S. S. Woodard, T. Katsuki, Y. Yamada, M. Ikeda, K. B. Sharp-
less, J. Am. Chem. Soc. 1981, 103, 6237–6240.
[19] Z. Li, B. T. Parr, H. M. L. Davies, J. Am. Chem. Soc. 2012, 134, 10942–
10946.
[10] For intramolecular examples, see: a) A. Aponick, B. Biannic, Org. Lett.
2011, 13, 1330–1333; b) A. Aponick, B. Biannic, M. R. Jong, Chem.
Commun. 2010, 46, 6849–6851; c) A. Aponick, C. Y. Li, B. Biannic, Org.
Lett. 2008, 10, 669–671; d) M. Bandini, M. Monari, A. Romaniello, M.
Tragni, Chem. Eur. J. 2010, 16, 14272–14277; e) B. Biannic, A. Aponick,
Eur. J. Org. Chem. 2011, 6605–6617; f) B. Biannic, T. Ghebreghiorgis, A.
Aponick, Beilstein J. Org. Chem. 2011, 7, 802–807; g) T. Ghebreghiorgis,
B. Biannic, B. H. Kirk, D. H. Ess, A. Aponick, J. Am. Chem. Soc. 2012, 134,
16307–16318; h) W. P. Unsworth, K. Stevens, S. G. Lamont, J. Robertson,
Chem. Commun. 2011, 47, 7659–7661.
[11] a) P. C. Young, N. A. Schopf, A.-L. Lee, Chem. Commun. 2013, 49, 4262–
4264; see also: b) P. Mukherjee, R. A. Widenhoefer, Chem. Eur. J. 2013,
19, 3437–3444; c) J. R. Wright, P. C. Young, N. T. Lucas, A.-L. Lee, J. D.
Crowley, Organometallics 2013, 32, 7065–7076.
[12] For example, see: a) K. Onitsuka, H. Okuda, H. Sasai, Angew. Chem.
2008, 120, 1476–1479; Angew. Chem. Int. Ed. 2008, 47, 1454–1457;
b) S. Ueno, J. F. Hartwig, Angew. Chem. 2008, 120, 1954–1957; Angew.
Chem. Int. Ed. 2008, 47, 1928–1931.
[13] For related gold-catalysed examples using other O- and N-nucleophiles,
see: a) G. Cera, M. Chiarucci, M. Bandini, Pure Appl. Chem. 2012, 84,
1673–1684; b) M. Chiarucci, M. Locritani, G. Cera, M. Bandini, Beilstein J.
Org. Chem. 2011, 7, 1198–1204; c) P. Kothandaraman, S. J. Foo, P. W. H.
Chan, J. Org. Chem. 2009, 74, 5947–5952; d) P. Mukherjee, R. A. Widen-
hoefer, Org. Lett. 2010, 12, 1184–1187; e) P. Mukherjee, R. A. Widenhoe-
fer, Org. Lett. 2011, 13, 1334–1337; f) P. Mukherjee, R. A. Widenhoefer,
Angew. Chem. 2012, 124, 1434–1436; Angew. Chem. Int. Ed. 2012, 51,
1405–1407.
[14] For example, see: a) S. Chandrasekhar, V. Jagadeshwar, B. Saritha, C. Nar-
sihmulu, J. Org. Chem. 2005, 70, 6506–6507; b) F. X. Felpin, Y. Landais, J.
Org. Chem. 2005, 70, 6441–6446; c) M. Frank, H. J. Gais, Tetrahedron:
Asymmetry 1998, 9, 3353–3357; d) H. J. Gais, T. Jagusch, N. Spalthoff, F.
Gerhards, M. Frank, G. Raabe, Chem. Eur. J. 2003, 9, 4202–4221; e) H. J.
Gais, N. Spalthoff, T. Jagusch, M. Frank, G. Raabe, Tetrahedron Lett. 2000,
41, 3809–3812; f) M. Jegelka, B. Plietker, Org. Lett. 2009, 11, 3462–3465;
g) T. Kondo, Y. Morisaki, S. Uenoyama, K. Wada, T. Mitsudo, J. Am. Chem.
Soc. 1999, 121, 8657–8658; h) N. Gao, X. Zhao, Eur. J. Org. Chem. 2013,
2708–2714.
[20] For a recent review, see: N. T. Patil, Chem. Asian J. 2012, 7, 2186–2194.
[21] M. Sasaki, H. Ikemoto, M. Kawahata, K. Yamaguchi, K. Takeda, Chem. Eur.
J. 2009, 15, 4663–4666.
[22] Chirality transfer is observed in the related allylic etherifications with
(S)-20 as a substrate.
[23] A. Zhdanko, M. Stroebele, M. E. Maier, Chem. Eur. J. 2012, 18, 14732–
14744.
[24] S. Gaillard, J. Bosson, R. S. Ramon, P. Nun, A. M. Z. Slawin, S. P. Nolan,
Chem. Eur. J. 2010, 16, 13729–13740.
[25] Calculations were run with the Gaussian suite of programs and em-
ployed the BP86 functional. Au and P centres described with the Stutt-
gart RECPs and associated basis set (with added d orbital polarisation
on P (z=0.387) and 6–31G** basis sets for all other atoms. Free ener-
gies are reported in the text, based the gas-phase values, incorporating
corrections for dispersion effects using Grimme’s D3 parameter set (i.e.,
BP86-D3) and solvent (CHCl3, PCM approach). See the Supporting Infor-
mation for references and full details.
[26] On the uncorrected SCF energy surface a two step process was charac-
terised via an intermediate related to species B in Scheme 3. However,
the subsequent proton transfer transition state fell to lower energy
than this intermedaite once the corrections for entropy, solvent and
dispersion effects were applied. This was generally the case for the SN2’
steps computed here, and in many cases no minimum corresponding
to structure B could be located. The SN2’ step therefore effectively be-
haves as a one-step, concerted process, as shown in Figures 2 and 4.
[27] A second SN2’ pathway was defined from IntI4’b in which the thiol at-
tacks syn to the metal. This entails a significantly higher barrier of
21.7 kcalmolꢀ1 and so would not be competitive with the anti attack
shown in the Figure 2. See the Supporting Information for details.
[28] We have also computed the [(Me3P)Au(SHPh)]+ adduct and found it to
have a relative free energy of +0.3 kcalmolꢀ1. This would therefore not
affect the overall computed barriers for the SN2’ reaction.
[29] H.-J. Gais, A. Bçhme, J. Org. Chem. 2002, 67, 1153–1161.
[30] The related gold(I)-catalysed isomerisation of allylic ethers is known,
see ref. [9b,d], and: a) R. S. Paton, F. Maseras, Org. Lett. 2009, 11, 2237–
2240; b) H. Wang, Q. He, K. Tan, Acta Chim. Sinica 2013, 71, 1663–1667.
[15] a) P. Jaisankar, S. Tanaka, M. Kitamura, J. Org. Chem. 2011, 76, 1894–
1897; b) M. Roggen, E. M. Carreira, Angew. Chem. 2012, 124, 8780–
8783; Angew. Chem. Int. Ed. 2012, 51, 8652–8655; c) S. Tanaka, P. K.
Received: April 28, 2014
Published online on July 30, 2014
Chem. Eur. J. 2014, 20, 11540 – 11548
www.chemeurj.org
11548 ꢀ 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim