Dynamic covalent chemistry (DCC)[1 ] has emerged in the last two decades as an area that combines the best attributes
of organic chemistry (synthesis of stable compounds) with those of supramolecular
chemistry (error correction).[2 ] At the heart of DCC are robust and reliable chemical reactions that, at least in
the presence of suitable catalysts,[3 ] lead to the formation of equilibrium mixtures under relatively mild conditions.
Typical examples include the exchange of disulfides with thiols,[4 ] and the reversible condensation reactions of imines,[5 ] hydrazones,[6 ] and oximes.[7 ] However, in the light of new applications of DCC, especially in the synthesis of
porous materials[8 ] and in the life sciences,[9 ] there is unabated interest in the development of new dynamic covalent reactions.
To this end, the groups of von Delius and of Furlan recently reported investigations
of O ,O ,O -orthoester exchange[10 ] and dithioacetal exchange reactions,[11 ] respectively (Scheme [1 ]). Both reactions share similarities, such as a requirement for acid catalysis in
organic solvents[12 ] and their widespread use in protective-group chemistry.[13 ] The tripodal nature of orthoesters makes them uniquely suited for the self-assembly
of cage-type architectures.[14 ]
[15 ] Also, orthoester exchange gives rise to a remarkable level of molecular diversity,
because by mixing one orthoester with one alcohol, an equilibrium mixture consisting
of four different orthoesters is obtained (Scheme [1 ]A). In contrast, dithioacetal exchange produces fewer products (Scheme [1 ]B) and is more suited to the preparation of cyclic hosts.[16 ] This exchange can be connected to that of disulfides and thioesters to generate
multilevel dynamic systems.[17 ] Compared with orthoesters, dithioacetals are less susceptible to hydrolysis and
they demand more-acidic media for exchange.
Scheme 1 Exchange reactions of (A) orthoesters, (B) dithioacetals, and (C) trithioorthoesters
In this study, we focus on trithioorthoester exchange, an area of chemical space that,
from the perspective of topology and reactivity, lies between orthoester exchange
and dithioacetal exchange (Scheme [1 ]C). Trithioorthoesters[18 ] can be obtained from, inter alia, O ,O ,O- orthoesters,[19 ] chloroform,[20 ] orthothioformates,[21 ] or dithioacetals,[22 ]
[23 ] and trithioorthoester groups have been used primarily as protective groups, especially
when the product is required to be more stable toward acid hydrolysis than its oxygen
counterpart.[13 ] Here, we present a comprehensive investigation of the conditions required for trithioorthoester
exchange and for trithioorthoester metathesis, in the hope that these transformations
will prove useful in areas where the advantages of thermodynamic control can be harnessed.[8e,24 ]
We started by testing the feasibility of trithioorthoester exchange under conditions
similar to those used for the activation of O ,O ,O -orthoesters[10 ] or dithioacetals.[11 ] To this end, trifluoroacetic acid (TFA; 1 or 10 equiv) was added to a chloroform-d solution containing tris(phenylthio)methane (A13
; 50 mM, 1 equiv) and phenylmethanethiol (2 ; 3 equiv) at room temperature (Table [1 ]). All solvents were dried over molecular sieves before use to minimize the irreversible
hydrolysis of trithioorthoesters to thioesters. The compositions of the reaction mixtures
were analyzed by 1 H NMR spectroscopy after one hour and 24 hours. The use of ten equivalents of TFA
led to equilibration of the mixture within one hour of reaction time. The apparent
bias toward trithioorthoesters rich in building block 2 (Table [1 ], entry 1) can be rationalized in terms of differences in steric demand. The undesired
hydrolysis reaction led to ≤2% of thioesters A1 and A2 with respect to the initial trithioorthoesters. A tenfold decrease in the amount
of TFA was possible, but led to slower exchange, as indicated by the dominance of
starting material A13
after 24 hours of reaction time (Table [1 ], entry 2). The harsh Brønsted acidic conditions, which are required to equilibrate
trithioorthoesters within a reasonable timespan, are similar to those giving rise
to dithioacetal exchange, whereas O ,O ,O -orthoester exchange proceeds under much milder conditions.
Table 1 Scope of Trithioorthoester Exchangea
Entry
Acid (Equiv)
Solvent
Reaction time (h)
A13
/A12 2 /A122
/A23
/(A1 + A2 )b
1
TFA (10)
CDCl3
1
3:17:38:42: –2:16:38:43:1
2
TFA (1)
CDCl3
1
97:3:–:–:–
3
TFA (10)
C6 D6
1
8:17:33:42:–
4
TFA (10)
CD3 CN
1
98:2:–:–:–
5
MsOH (1)
CD3 CN
1
97:3:–:–:–
6
PTSA (1)
CD3 CN
1
60:24:1:5:–2:11:37:50:–
7
H2 SO4 (1 )
CD3 CN
1
73:18:6:3:–4:14:39:43:–
8
TfOH (1)
CD3 CN
1
1:18:42:39:–
9
FeCl3 (0.1)
CD3 CN
1
35:26:23:16:–3:15:39:43:–
10
AlCl3 (0.1)
CD3 CN
1
92:8:–:–:–14:15:29:42:–
11
BF3 ·OEt2 (0.1)
CD3 CN
1
78:18:4:–:–9:14:33:44:–
12
FeCl3 (0.1)
CDCl3
1
3:17:40:40:–
a Reaction conditions: A13
(37.5 μmol), 2 (112.5 μmol), TFA (375 μmol), CDCl3 (V
total : 750 μL), r.t.
b Product ratios were determined by integration of 1 H NMR signals. Equilibrium mixtures are highlighted in bold face; ‘–’ indicates not
detected. For full spectra, see Figures S4–S25 in the Supplementary Information.
A range of parameters was next investigated to better understand the reaction. First,
various solvents were tested by using TFA in a standard amount of ten equivalents.
In benzene-d
6 , equilibrium was reached after one hour of reaction (Table [1 ], entry 3), whereas in acetonitrile-d
3 the exchange was slow and the composition after 24 hours was still far from equilibrium
(Table [1 ], entry 4). No exchange products were observed in DMSO-d
6 or THF-d
8 . When stronger Brønsted acids were compared in acetonitrile-d
3 (for solubility reasons), a correlation between the pK
a value[25 ] and the reaction kinetics was observed. Slow exchange was observed with TFA (pK
a = 12.7) or methanesulfonic acid (pK
a = 10.0) (entries 4 and 5), whereas the mixtures with p -toluenesulfonic acid (pK
a = 8.0) or sulfuric acid (pK
a = 7.2) equilibrated after 24 hours (entries 6 and 7). Only the mixture with trifluoromethanesulfonic
acid (pK
a = 2.6) equilibrated after one hour of reaction (entry 8). Trifluoromethanesulfonic
acid turned out to be a less effective catalyst in DMSO-d
6 , whereas THF-d
8
[26 ] was unstable in the presence of this acid [see the Supplementary Information (SI)].
Stoichiometric or excess Brønsted acids were shown to be useful for promoting trithioorthoester
exchange. These conditions were used in exchange experiments initiated from various
starting materials,[1c ]
[27 ] which confirmed the reversibility of the reaction (Figures S1–S3, SI).
In the hope of identifying milder and truly catalytic (substoichiometric) conditions,
we proceeded to investigate some representative Lewis acids.[28 ] For reasons of solubility, we chose CD3 CN as a solvent to investigate FeCl3 , AlCl3 , and BF3 ·OEt2 , whereas CDCl3 was used in combination with SnCl4 , TiCl4 , and FeCl3 . We found that one equivalent of each of the Lewis acids FeCl3 , AlCl3 , SnCl4 , and BF3 ·OEt2 led to equilibration after roughly one hour; experiments with 0.1 equivalent of these
Lewis acids showed that they can indeed be regarded as catalysts for this reaction
(Table [1 ], entries 9–12) with the following reactivity trend: FeCl3 > BF3 ·OEt2 > AlCl3 . In addition, the use of Lewis acids in CDCl3 was found to be promising. When 0.1 equivalent of FeCl3 in CDCl3 was tested, equilibrium was attained after one hour (entry 12). Although the equilibrium
position seems to be unaffected by the presence of the iron cation, signal broadening
was observed, which increased with reaction time, preventing accurate integration
after 24 hours of reaction. These results show that Lewis acids require a lower catalyst
loading than do Brønsted acids. Future work might focus on Lewis acid catalysis in
CDCl3 to achieve shorter reaction times when using substoichiometric amounts of catalyst.
To this end, alternative analytical tools such as HPLC might be helpful in avoiding
interference by the paramagnetic effects induced by metal cations.
To explore the scope of the reaction, we studied the effects of the nucleophile on
the kinetics and thermodynamics of exchange. To this end, trithioorthoester A13
was combined with thiols 2 –6 in the presence of TFA (10 equiv) (Scheme [2 ]). Primary thiols such as phenylmethanethiol (2 ) or hexane-1-thiol (3 ) gave compositions shifted toward trithioorthoester products containing more-exchanged
units. Secondary and aromatic thiols, such as propane-2-thiol (4 ) and 4-methylbenzenethiol 5 , respectively, led to nearly symmetrical statistical distributions of trithioorthoesters.
The use of bulky adamantane-1-thiol (6 ) led to the formation of only a single exchange product containing one building block
6 . The exchange kinetics were not noticeably affected by thiol structures, because
all the mixtures were equilibrated after one hour of reaction time. Comparable equilibrium
distributions were obtained after 24 hours when A13
was exposed to the nucleophiles in CD3 CN in the presence of FeCl3 (0.1 equiv) as catalyst (see SI). These results are in agreement with those observed
in O ,O ,O -orthoester exchange, in which the equilibrium position, but not the equilibration
kinetics, depends on the size of the nucleophile.[10 ] Finally, inspired by recent studies on (crossed) dichalcogenide exchange reactions,[29 ] we investigated the reaction of trimethyl trithioorthoformate A83
with benzeneselenol 7 . After one hour, four signals in the diagnostic trithioorthoester range were observed;
these appeared in a statistical proportion, showing the feasibility and the reversibility
of the reaction (after 24 hours, selenol oxidation dominates the reaction outcome).
To our knowledge, this is the first example of selenol/trithioorthoester exchange.
Scheme 2 Effect of the structure of the nucleophile on the distribution of trithioorthoesters
at equilibrium. Reaction conditions : A13
or A83
(37.5 μmol), nucleophile (112.5 μmol), TFA (375 μmol), CDCl3 (V
total : 750 μL), r.t. Product ratios were determined by integration of 1 H NMR signals after 1 h. Hydrolysis was negligible (<2%) in all cases except for thiol
6 (7%). a Exchange performed with A13
(CDCl3 or CD3 CN). b Exchange performance with A83
(CDCl3 ). For full spectra, see Figures S26–S36 (SI).
While keeping the ratio of trithioorthoester A13
to thiol 2 at 1:3, and with a constant total amount of TFA, we next investigated the effect
of the trithioorthoester concentration (Figure S37). We found that a tenfold decrease
in the concentration of A13
from 50 mM to 5 mM did not affect the equilibrium position. This indicates that trithioorthoester
exchange might well be suited to experiments requiring low concentrations of reactants.
However, a simultaneous decrease in the concentration of TFA to 10 equivalents (50 mM)
slowed the exchange reaction considerably (96% of A13
was present after 1 h).
To take advantage of the low boiling point of methanethiol (6 °C), A83
was treated with 4-methylbenzenethiol (5 ) in the presence of excess TFA (10 equiv) under a smooth nitrogen stream while the
temperature was increased from r.t. to 40 °C and then to 60 °C to shift the equilibrium
completely towards trithioorthoesters containing exchanged side chains.[30 ] The composition could indeed be shifted almost completely toward the formation of
exchange product A53
(Figure S38). This straightforward method might be useful whenever a complete shift
in the equilibrium toward the product side is desired.
In a comparative study, we examined the hydrolytic stability of trimethyl trithioorthoformate
[(S ,S ,S )-A83
], its oxygen-containing counterpart trimethyl orthoformate [(O ,O ,O )-A93
], and the dithioacetal bis(methylthio)methane (AH82
).[31 ] Increasing amounts of TFA were added to solutions containing each compound with
one equivalent of water, and the samples were analyzed by 1 H NMR spectroscopy after one hour of reaction (Figure [1 ]; note the logarithmic scale of the x -axis).
Figure 1 Effects of increasing amounts of TFA on the hydrolysis of orthoester (O ,O ,O )-A93
, trithioorthoester (S ,S ,S )-A83
, and dithioacetal AH82
. Reaction conditions : A93
or A83
or AH82
(37.5 μmol), increasing amounts of TFA [from 37.5 nmol (50 μM) to 3.75 mmol (5.0 M)],
CDCl3 (V
total : 750 μL), r.t., 1 h; internal standard: toluene. For full spectra, see Figures S39–S56
in the SI.
As expected, the trithioorthoester compound was considerably more stable to hydrolysis
than was the orthoester, but was less stable than the dithioacetal. We found that
the addition of ten equivalents of TFA led to complete hydrolysis of (O ,O ,O )-A93
, whereas (S ,S ,S )-A83
and AH82
were mostly stable. The addition of 100 equivalents of TFA was necessary to hydrolyze(S ,S ,S )-A83
completely, whereas AH82
remained stable under these conditions. These findings indicate that in terms of
hydrolytic stability, the trithioorthoester compound is located at an intermediate
position between the orthoester and dithioacetal compounds. Most importantly, the
results point towards faster exchange versus slower hydrolysis kinetics of trithioorthoesters:
over a one hour period, treatment of (S ,S ,S )-A83
with ten equivalents of TFA led to complete equilibration (Figure S57), but only
to 30% hydrolysis. This observation explains why the rigorous exclusion of water is
not as essential for (S ,S ,S )-orthoester exchange as it is for the (O ,O ,O )-orthoester variant. When trimethyl trithioorthoacetate (C83
) was hydrolyzed with ten equivalents of TFA, the hydrolysis ratio after one hour
was 93%, indicating that the same electronic effects as previously reported account
for differences in hydrolytic stability (Figure S58).[15e ] The observed differential susceptibility to hydrolysis of orthoesters, trithioorthoesters,
and dithioacetals is relevant for the selective removal of protective groups.
In light of the shuttle catalysis concept,[32 ] there has recently been an increase in interest in exchange reactions between molecules
having the same kind of functional group, i.e. Type 1 metathesis.[33 ] Among the Type 1 metathesis reactions that have been studied from the perspective
of dynamic covalent/combinatorial chemistry are disulfide,[4 ]
[34 ] trithiocarbonate,[35 ] thiazolidine,[36 ] acetal,[37 ] orthoester,[10 ] and dithioacetal exchange.[17a ] We therefore wondered whether a direct metathesis reaction between two trithioorthoesters
might be possible.
To answer this question, a set of trithioorthoesters suitable for crossover experiments
were synthesized.[38 ]
[39 ]
[40 ]
[41 ] Treatment of compounds A13
and B83
with TFA (10 equiv) led to a Type I metathesis, as indicated by the appearance in
the 1 H NMR spectrum of a set of signals corresponding to the eight expected trithioorthoesters
(Figure [2 ] and Table [2 ], entry 1). As in the case of related orthoester metathesis,[10 ] we believe that the generation of small quantities of free thiol[42 ] is responsible for the observed reactivity.
Figure 2 Trithioorthoester metathesis between A13
and B83
(top) and 1 H NMR spectrum measured after one hour (bottom). The two dashed boxes show the corresponding
formate and acetoxy hydrogen atoms of trithioorthoesters formed from A13
and B83
. Reaction conditions : A13
(37.5 μmol), B83
(37.5 μmol), TFA (375 μmol), CDCl3 (V
total : 750 μL), r.t., 1 h. Product distributions are given by normalized integral values
below the corresponding peaks for both sets. See also Table [2 ], entry 1 and the full spectra in Figure S59 in the SI.
Table 2 Metathesis between Different Trithioorthoesters
Entry
Substrate 11
3
Substrate 22
3
Reaction outcome1
3 /HR1
2 R2 /HR1 R2
2 /HR2
3 /thioesters (hydrolysis)a
1b
A13
B83
17:43:33:7:–
2c
A13
A53
13:36:36:13:2
3c
A53
A(10)3
13:35:36:13:3
4d
A13
A83
10:36:37:11:6
5b
A13
A(11)3
12:34:33:10:12
6c
A53
A43
18:28:30:12:12
7b
A(11)3
C13
3:21:42:28:6
8b
A(11)3
D13
ether cleavage
a Product ratios were determined by integration of 1 H NMR signals after 1 h of reaction; ‘–’ indicates not detected. For full spectra,
see Figures S60–S67 in the SI.
b Reaction conditions: HR1
3
(37.5 μmol), XR2
3
(37.5 μmol), TFA (375 μmol), CDCl3 (V
total : 750 μL), r.t.
c Reaction conditions: HR1
3
(15 μmol), XR2
3
(15 μmol,), H2 SO4 -saturated CDCl3 , (V
total : 500 μL), r.t.
d Reaction conditions: HR1
3
(37.5 μmol), XR2
3
(37.5 μmol), FeCl3 (3.75 μmol), CD3 CN (V
total : 750 μL), r.t.
To explore the generality of this finding, additional metathesis reactions were carried
out with other pairs of trithioorthoesters. Combinations of trithioorthoesters containing
different aromatic substituents on sulfur (Table [2 ], entry 2) or aromatic and primary alkyl substituents (entries 3–5) led to statistical
distributions, whereas trithioorthoesters containing aromatic and secondary thiol
side chains (entry 6) led to a slightly biased composition favoring the starting aromatic
trithioorthoester. These experiments can also be regarded as competitive hydrolysis
experiments: throughout our investigations, we found that hydrolysis of trithioorthoesters
with alkyl residues on sulfur is faster than that for aromatic ones (compare Figure
S60 with Figures S61–S64; SI). Next, we investigated the influence of electron-withdrawing
and electron-donating substituents on trithioorthoester metathesis by conducting metathesis
reactions between tris(ethylthio)methane [A(11)3
] and trithioorthoesters C13
and D13
, containing electron-withdrawing and electron-donating groups, respectively (Figures
S65 and S66). After one hour, similar equilibrium distributions for the formate species
and a comparable degree of hydrolysis were observed in both cases (entries 7 and 8).
However, in the case of D13
, degradation of the starting material, presumably by ether cleavage[43 ] occurred, representing a notable limitation of this method. Finally, several attempts
were made to carry out crossed metathesis reactions between (S ,S ,S )-A13
and (O ,O ,O )-A93
; these were unsuccessful due to instantaneous hydrolysis of O ,O ,O -orthoester, further confirming the previously observed differences in stability (Figure
S67).
In summary, we have described a methodological investigation of the exchange reaction
between trithioorthoesters and thiols, as well as the direct trithioorthoester metathesis
reaction. These reactions have two appealing properties in the context of other reversible
covalent reactions. First, the tripodal structure of S ,S ,S -orthoesters provides an elegant entry to sulfur-rich three-dimensional architectures
with possible applications in the removal of heavy-metal ions. Second, the harsh conditions
necessary to initiate these exchange reactions might be advantageous for applications
in materials science, such as the (solvothermal) synthesis of porous materials.
