Adamantane exhibits fascinating properties, such as rigidity, lipophilicity, steric
bulkiness, electron richness, and chemical inertness.[1] By virtue of these desirable features diamondoids, i.e., diamond-like, nanometer-sized
aliphatic cage hydrocarbons, for which adamantane is the parent, are used as catalyst
backbones,[2] dispersion energy donors,[3] or molecular rectifiers.[4] Adamantane is an attractive moiety in the design of drugs, e.g., memantine and saxagliptin
(both belonging to the top 200 drugs by worldwide sales 2016),[5] favorably affecting administration, distribution, metabolism, and excretion (ADME)
properties.[6] The selective functionalization of diamondoids is essential for their application,
and this presents a formidable challenge in selective sp3-C–H bond functionalization. Usually hydroxylation, bromination, or amination (via Ritter reaction) are used as the first step.[1] The resulting compounds can then readily be functionalized further. However, due
to the chemical inertness of unactivated aliphatic C–H-bonds, functionalization requires
harsh conditions, making the synthesis of orthogonally disubstituted adamantanes challenging
(Scheme [1]). A mild direct C(sp3)–H functionalization of monosubstituted adamantanes would greatly streamline the
synthesis of these diamondoids and facilitate their use even more. The incorporation
of nitriles is particularly desirable, as they are widely present as key functional
groups in a variety of natural products and pharmaceuticals.[7] They can readily be converted into carboxylic acids, esters, amines, or amides.
Furthermore, they can be used in [3+2] cycloadditions affording heterocycles such
as tetrazoles and oxadiazolines (Scheme [1]).[8]
Scheme 1 Orthogonally bifunctionalized adamantane derivatives
Due to the versatility of nitriles, a variety of cyanation methods were developed.
While C(sp2)–H cyanations are well established,[9] C(sp3)–H cyanations remain challenging. They are often limited to the synthesis of pre-functionalized
precursors such as alkyl iodides,[10] allylic compounds,[11] or enolates.[12] Recently, two photocatalytic systems for C(sp3) cyanation were reported utilizing alkyltrifluoroborates[13] and carboxylates[14] as precursors. Additionally, Sun and coworkers reported an oxidative or free radical
C(sp3)–H cyanation for alkanes, ethers, and tertiary amines.[15] Moreover, cyanations of activated C(sp3)–H bonds, such as α-heteroatoms[16] and an enantioselective benzylic C–H cyanation[17] have been established. Although some of these methods tolerate a broad range of
functional groups, their use is restricted due to elaborate precursor synthesis. Direct
cyanations of unactivated C(sp3)–H bonds are rare. Only a few direct cyanations exist, including the pioneering studies
by Müller and Huber in 1963, who introduced an unselective cyanation with cyanogen
chloride[18] and a photoexcited benzophenone mediated C(sp3)–H cyanation.[19] While the latter approach is applicable to benzylic and aliphatic C–H bonds, the
differentiation of these bonds is only modest, due to the use of benzophenone as the
hydrogen atom transfer (HAT) reagent. Furthermore, aliphatic substrates often require
the use of excess of substrate.
In continuation of our work with phthalimido-N-oxyl (PINO),[20] we envisioned a selective C(sp3)–H cyanation of adamantanes using N-hydroxyphthalimide (NHPI).[21] To the best of our knowledge, PINO has never been used in aliphatic C–H bond cyanations.[22] We envisioned a three-step process, consisting of the formation of PINO from its
precursor NHPI, followed by HAT from adamantane 1 to PINO, generating an adamantyl radical 1•
, which is trapped by p-toluenesulfonyl cyanide (TsCN), thereby affording the desired cyanated product 2 (Scheme [2]).
Scheme 2 PINO-catalyzed C(sp3)–H cyanation concept
We commenced our studies with 1 as a model substrate and p-toluenesulfonyl cyanide (TsCN) as the electrophilic cyanide source[23] (Table [1]). At first we performed an initial screening to elaborate suitable reaction conditions
for the generation of PINO. Systems such as α,α′-azobisisobutyronitrile (AIBN), cobalt
(II/III) salts/O2, and cerium(IV) ammonium sulfate (CAS) afforded only low yields of 1-cyano adamantane
2
[24] and significant amounts of starting material 1 remained. However, cerium(IV) nitrate (CAN)[25] was shown to be superior, resulting in 42% yield of 2. On the other hand, the use of CAN also led to the formation of significant amounts
of 1-nitro adamantane 3
[26] (13%). This side product forms by reduction of Ce(IV) → Ce(III), thus generating
HNO3.[22] HNO3 itself is capable to generate PINO, thereby releasing •NO2 that recombines with the intermittently formed 1-adamantyl radical.[27] This pathway was confirmed with the use of 1 equiv HNO3, affording 15% of 1-nitro adamantane 3.
Table 1 Screening of Systems for the PINO-Catalyzed Cyanationa
|
|
SET reagent
|
2 (%)b
|
|
AIBNc
|
5
|
|
Co(acac)3
d
|
7
|
|
Co(OAc)2
d
|
8
|
|
CANe
|
42
|
|
CASe
|
9
|
|
HNO3
e
|
47
|
a Reaction conditions: 0.5 mmol scale, ratio of1/NHPI/TsCN (1:0.1:2), 5 mL 1,2-dichloroethane, 16 h, 75 °C.
b Yields determined by GC with hexadecane as internal standard.
c 3 mol% AIBN.
d 1 mol% of the corresponding metal salt, 1 atm air.
e 1 equiv.
We envisioned to suppress the formation of 3 by capturing the released HNO3 (Table [2]). Initial screening of bases including MgO, acetates, and carbonates, showed that
carbonates performed best. The use of Cs2CO3 afforded no product, while Ag2CO3, Na2CO3, and Li2CO3 led to an increase of the yield with up to 77% for Li2CO3. In general, the yield increased with decreasing cation radius; 1 equiv Li2CO3 worked best. Furthermore, we tested Co(acac)3, a known cocatalyst[28] for PINO, but it was ineffective under the chosen conditions. Importantly, the cyanation
selectively proceeds at the tertiary C–H position, thus indicating that PINO performs
a chemoselective hydrogen abstraction due to its polarity.[21b]
Table 2 Influence of Inorganic Basesa
|
|
Base (1 equiv)
|
2 (%)b
|
3 (%)b
|
|
Cs2CO3
|
0
|
0
|
|
Ag2CO3
|
56
|
<5
|
|
Na2CO3
|
73
|
7
|
|
Li2CO3
|
77
|
5
|
|
Li2CO3 (1.5 equiv)
|
47
|
<5
|
|
Li2CO3 (0.5 equiv)
|
51
|
11
|
|
Na2CO3
c
|
71
|
8
|
|
Na2CO3
c,d
|
60
|
14
|
a Reaction conditions: 0.5 mmol scale, ratio of 1/NHPI/CAN/TsCN/base (1:0.2:1:2:1), 5 mL 1,2-dichloroethane, 16 h, 75 °C.
b Yields determined by GC with hexadecane as internal standard.
c 0.1 equiv NHPI.
d 1 mol% Co(acac)3.
With the optimized conditions in hand, we focused on the cyanation of various substrates
(Scheme [3]). Common C–H activation procedures require excess of the starting material, in order
to suppress a second C–H activation of the product. Under our conditions, the cyanation
requires only 1 equiv starting material, thereby affording, e.g., 1-cyano adamantane
2 in 69% yield without formation of the dicyanated product. The strong electron-withdrawing
cyano group deactivates the cage, thus suppressing subsequent cyanation. This was
illustrated by utilizing 2, affording 1,3-dicyano adamantane 4
[29] in only 20% yield. Hence, only 1 equiv substrate is necessary in the cyanation.
Methyl adamantane (5)[30] was isolated in 71% yield, while dimethyl 6
[31] and trimethyl 7
[32] substituted adamantanes were isolated in 33% and 28% yield, respectively. The same
reactivity trend for the corresponding halogenated derivatives was observed by Olah
and coworkers in a Lewis acid catalyzed cyanation.[33] The di- and trisubstituted adamantanes were cyanated in comparable yields to the
corresponding methyl derivatives, affording cyanated products 8
[34] and 9.[35] The cyanation of 1-bromo adamantane afforded product 10
[36] in 25% yield, with traces of N-tosyloxyphthalimide S1,[37] formally a product of the radical recombination of tosylsulfonyl and a PINO radical.
Upon addition of 20 mol% NHPI after 6 h the yield of 10 increased to 34%. 1-Cyano-3-phenyl adamantane 11
[38] was obtained in 47% and the alkynyl derivative 12
[39] in 41% yield. Notably, 12 represents an orthogonal building block, containing a triple bond, which is otherwise
not easily accessible. Adamantane carboxylic acid methyl ester was cyanated in 50%
yield (13).[40] Silyl-protected alcohols can be also used under the chosen conditions, affording
39% (14a).[41] Use of phthalimides, acetamides, and azides in the cyanation results in the formation
of γ-amino acid derivatives 15–17.[42] Note that the use of 17 results in another orthogonally difunctionalized building block readily available
for ‘click-reactions’. Furthermore, diamantane can be cyanated in 60% yield, affording 18
[43] in a ratio of 4.3:1 in favor of the medial position. The cyanation of 4-diamantane
carboxylic acid methyl ester afforded 35% yield in a ratio of 1:1.1 (19m1
:19m2
).[44] Particular the syntheses of these cyanated products 19 would require considerably more steps in comparison to established methods.
Scheme 3 Substrate scope of the PINO-catalyzed C(sp3)–H cyanations. Yields of isolated, pure products are given. Reaction conditions: 0.5 mmol starting material, ratio of SM/NHPI/CAN/TsCN/base (1:0.2:1:2:1), 5 mL 1,2-dichloroethane,
16 h, 75 °C. a 20 mol% NHPI added after 6 h. b Ratio determined by GC-MS.
Table 3 Mechanistic Investigationsa
|
|
Conditions
|
2 (%)b
|
|
NaCN (2 equiv), TBACN (0.15 equiv)
|
traces (< 1%)
|
|
I2 (1 equiv)
|
0
|
|
0 mol% NHPI
|
10
|
|
dark
|
40
|
|
365 nm
|
37
|
a Reaction conditions: 0.5 mmol scale, ratio of 1/NHPI/CAN/TsCN/base (1:0.2:1:2:1), 5 mL 1,2-dichloroethane, 16 h, 75 °C.
b Yields determined by GC with hexadecane as internal standard.
In order to support our initial mechanistic hypothesis of a radical pathway, several
tests were performed (Table [3]). In the presence of a radical scavenger such as I2, no product formed. The isolation of traces of N-tosyloxyphthalimide S1, a result of a radical recombination, supports this hypothesis. By performing a control
experiment with the optimized conditions (1 equiv TsCN, 1 equiv Li2CO3, 0.2 equiv NHPI, 75 °C, 16 h) the formation of S1 by a nucleophilic substitution could be excluded. In addition, the absence of NHPI
afforded only 10% of the cyanated product. This underscores that PINO indeed is the
catalytically active species. The exclusion of light decreased the yield to 40% yield.
CAN is known to produce nitroxyl radicals upon UV irradiation, while CeIV is reduced to CeIII.[45] However, irradiation of the reaction mixture at 365 nm afforded only 37% yield of
2. This result may indicate that low concentrations of nitroxyl radicals facilitate
the cyanation, while a higher concentration of •NO3 leads to a higher probability of termination events, consequently lowering the yield.
Furthermore, CAN is reduced upon UV irradiation, thus it is not available for the
generation of PINO and finally affording a lower yield. Moreover, the oxidation of
the adamantyl radical 1•
to the cation[25]
[46] could be excluded by the use of a CN− source (NaCN) in combination with a phase-transfer catalyst (TBACN), affording less
than 1% yield.
In summary, we report a novel direct C(sp3)–H cyanation of adamantane and two diamantane derivatives, utilizing only 1 equiv
of substrate.[47] The method allows the efficient synthesis of substituted cyano adamantanes. A variety
of these valuable compounds was synthesized for the first time. Mechanistic experiments
support a radical mechanism.
