.
Angewandte
Communications
similar, if not identical, signal shifts would be expected. In
contrast, 1H·22+·1H (Figure 3c) is clearly present. From the 10
possible pseudorotaxane structures, the head-to-head homo-
pseudo[3]rotaxane (H,H)-1H·22+·1H (Figure 1) emerged as the
only product (Figure 3). This result is in good agreement with
the total binding constants (Ktot = K1·K2)[13] as evaluated by
integration of the 1H NMR signals of the pseudorotaxanes in
To expand the code, we synthesized axle 42+ with two
dibenzylammonium stations.[19] First, we tested the binding of
the new axle to the tert-butylated calix[6]arene 1tBu. The
1H NMR spectrum of a 1:2 mixture of 42+ and 1tBu in CDCl3
(see Figure S22) reveals shielded benzylic resonances in the
4–6 ppm region, whereas no signals for shielded p-alkoxy-
benzyl moieties were found. Thus, double endo-benzyl
complexation of 42+ to yield the pseudo[3]rotaxane (T,T)-
the [1+1] mixtures ((H,H)-1H·22+·1H: Ktot = 5.8 Æ 0.3 ꢀ 106 mÀ2
;
(T,T)-1tBu·22+·1tBu
:
K
tot = 2.5 Æ 0.4 ꢀ 105 mÀ2). The 23-fold
1
tBu·42+·1tBu (Figure 1) was observed (Ktot = 1.8 Æ 0.3 ꢀ
higher binding constant of (H,H)-1H·22+·1H clearly renders it
the dominant product.
104 mÀ2).[19] Twofold complexation of the wheel was confirmed
by the presence of the 1tBu·42+·1tBu dication at m/z 1298 in the
ESI mass spectrum (see Figure S23).[19] Regarding the binding
ability of 42+ toward non-tert-butylated calix[6]arene 1H, the
1H NMR spectrum of a 1:2 mixture of 42+ and 1H (see
Figure S28a) showed shielded p-alkoxybenzyl resonances
(AX systems) in the 4–6 ppm region. This result was
confirmed by a COSY spectrum, which indicated the presence
of three pseudo[3]rotaxanes with a slight preference for endo-
p-alkoxybenzyl complexation.
ESI mass spectra obtained for a 1:2:2 mixture of 22+, 1H,
and 1tBu indicate a 17:9:1 ratio for 1H·22+·1H (m/z 898),
1H·22+·1tBu (m/z 1066), and 1tBu·22+·1tBu (m/z 1234; see Fig-
ure S26 in the Supporting Information). The deviation from
the NMR spectroscopic results and with it the seemingly
imperfect self-sorting may be attributed to the different
conditions during the MS experiment.[19]
When the heteropseudo[3]rotaxane dication 1H·22+·1tBu
was mass-selected and subjected to infrared multiphoton
dissociation (IRMPD, Figure 3d), 22+·1tBu (m/z 706) was more
prominent than 22+·1H (m/z 537). As the experiment probes
the activation energies for the two possible dissociation
processes, we can conclude that 1tBu has a higher kinetic
barrier to deslipping than 1H.[21]
In a [1+2] self-sorting experiment, axle 42+ and wheels 1H
and 1tBu were used in a 1:2:2 ratio in CDCl3. The correspond-
ing 1H NMR spectrum (see Figure S28b) clearly shows endo-
benzyl resonances corresponding to (T,T)-1tBu·42+·1tBu in the
4–6 ppm region and rules out the presence of other homo- or
heteropseudorotaxanes.
The above results yielded the first rules of our molecular
code (Figure 4). Rule 1 (social behavior): Benzylalkylammo-
nium sites on diammonium axles (e.g. 22+) select non-tert-
On the basis of this result, the molecular code could now
be expanded with two new rules (Figure 4). Rule 3 (social
behavior): Benzyl-p-alkoxybenzylammonium sites on a dia-
mmonium axle (e.g. 42+) select tert-butylated calix[6]arenes
(e.g. 1tBu) over non-tert-butylated calix[6]arenes (e.g. 1H),
which remain uncomplexed. Rule 4 (stereochemistry):
Threading of a benzyl-p-alkoxybenzylammonium site (e.g.
in 42+) through a tert-butylated hexaalkoxycalix[6]arene (e.g.
1
tBu) occurs with preference for the endo-benzyl configura-
tion.
On the basis of the complete set of all four rules,[22] we
could now probe the validity of the code in more complex
systems. Therefore, [2+2] self-sorting experiments with two
different axles and two different wheels were performed. As
shown in Figure 2 (blue box), a total of 20 possible homo- and
heteropseudo[3]rotaxanes can form with symmetrical axles.
In a [2+2] self-sorting experiment, axles 22+ and 42+ and
wheels 1H and 1tBu were used in a 1:1:2:2 ratio in CDCl3. The
corresponding 1H NMR spectrum (see Figure S30b) was
clearly the superposition of the spectra for the individual
pseudo[3]rotaxanes (H,H)-1H·22+·1H and (T,T)-1tBu·42+·1tBu
(see Figure S30a,c), whereas alternative combinations of
axles and wheels were not found. The formation of these
pseudo[3]rotaxanes was evident from the endo-benzyl reso-
nances corresponding to (T,T)-1tBu·42+·1tBu in the 4–6 ppm
region and the characteristic endo-alkyl resonances for
(H,H)-1H·22+·1H at negative ppm values. As above, the ESI
mass spectra did not show a perfect self-sorting picture, but
the trend towards the situation expressed in the NMR spectra
was clearly observed (see Figure S31). These results clearly
confirm the validity of the molecular code even for the more
complex [2+2] system, in which only two out of 20 possible
pseudo[3]rotaxanes formed stereospecifically. Further experi-
ments with different [2+2] systems corroborated a wider
Figure 4. The “molecular code” with respect to both benzylalkylammo-
nium and benzyl-p-alkoxybenzylammonium sites.
butylated calix[6]arene wheels (e.g. 1H) in preference to tert-
butylated wheels (e.g. 1tBu), which thus remain uncomplexed.
Rule 2 (stereochemistry): Threading of a directional alkyl-
benzylammonium station (e.g. in 22+) through a non-tert-
butylated hexaalkoxycalix[6]arene (e.g. 1H) occurs with
preference for the endo-alkyl orientation.[13]
These rules were corroborated by a second [1+2] sorting
system, in which the symmetrical bis(benzylalkylammonium)
axle 32+ (Figure 1), with terminal alkyl chains, was mixed with
two equivalents of each of the wheels 1H and 1tBu. As
predicted by the rules, 1H was selected also by axle 32+, so that
1H·32+·1H formed exclusively out of the 10 possible isomers
with the expected T,T configuration (see Figure S27)—again
in line with the total binding constants[13] ((T,T)-1H·32+·1H:
K
tot = 7.4 Æ 0.2 ꢀ 105 mÀ2; (T,T)-1tBu·32+·1tBu: Ktot = 2.2 Æ 0.3 ꢀ
103 mÀ2).
4
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 1 – 6
These are not the final page numbers!