Photoswitchable dihydroazulene macrocycles for solar energy storage: The effects of ring strain

Article abstract of DOI:10.1021/acs.joc.7b01760

Efficient energy storage and release are two major challenges of solar energy harvesting technologies. The development of molecular solar thermal systems presents one approach to address these issues by tuning the isomerization reactions of photo/ thermos

Full text of DOI:10.1021/acs.joc.7b01760

Subscriber access provided by UNIVERSITY OF ADELAIDE LIBRARIES
Article
Photoswitchable Dihydroazulene Macrocycles for
Solar Energy Storage: The Effects of Ring Strain
Alexandru Vlasceanu, Benjamin Normann Frandsen, Anders B. Skov, Anne Schou Hansen, Mads
Georg Rasmussen, Henrik Grum Kjaergaard, Kurt V. Mikkelsen, and Mogens Brøndsted Nielsen
J. Org. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.joc.7b01760 • Publication Date (Web): 30 Aug 2017
Downloaded from http://pubs.acs.org on August 30, 2017
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
The Journal of Organic Chemistry is published by the American Chemical Society.
1155 Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 13
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
Photoswitchable Dihydroazulene Macrocycles for Solar Energy Stor-
age: The Effects of Ring Strain
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Alexandru Vlasceanu, Benjamin N. Frandsen, Anders B. Skov, Anne Schou Hansen, Mads
Georg Rasmussen, Henrik G. Kjaergaard, Kurt V. Mikkelsen, Mogens Brøndsted Nielsen*
Department of Chemistry, Center for Exploitation of Solar Energy, University of Copenhagen, Universitetsparken
5, DKꢀ2100 Copenhagen Ø, Denmark. Eꢀmail: mbn@chem.ku.dk
Keywords: electrocyclic reactions, energy conversion, macrocycles, photochromism, strained molecules
ABSTRACT: Efficient energy storage and release are two major challenges of solar energy harvesting technologies.
The development of molecular solar thermal systems presents one approach to address these issues by tuning the
isomerization reactions of photoꢀ/thermoswitches. Here we show that the incorporation of photoswitches into macrocyꢀ
clic structures is a particularly attractive solution for increasing the storage time. We present the synthesis and properꢀ
ties of a series of macrocycles incorporating two dihydroazulene (DHA) photoswitching subunits, bridged by linkers of
varying chain length. Independent of ring size, all macrocycles exhibit stepwise, lightꢀinduced, ringꢀopening reactions
(DHAꢀDHA to DHAꢀVHF to VHFꢀVHF; VHF = vinylheptafulvene) with the first DHA undergoing isomerization with a
similar efficiency as the uncyclized parent system while the second (DHAꢀVHF to VHFꢀVHF) is significantly slower. The
energyꢀreleasing, VHFꢀtoꢀDHA, ring closures also occur in a stepwise manner and were systematically found to proꢀ
ceed slower in the more strained (smaller) cycles, but in all cases with a remarkably slow conversion of the second
VHF to DHA. We managed to increase the halfꢀlife of the second VHFꢀtoꢀDHA conversion from 65 h to 202 h at room
temperature by simply decreasing the ring size. A computational study reveals the smallest macrocycle to have the
most energetic VHFꢀVHF state and hence highest energy density.
Introduction
no CO2 emissions) as only the DHAꢀtoꢀVHF conversion
is induced by light. Light harvesting in this manner genꢀ
erates the metaꢀstable VHF isomer, which represents an
energy reservoir that discharges in time.
Molecular photoswitches are attractive entities for the
development of advanced functional nanostructures and
materials such as optical data storage devices, logic
gates, and sensors, as well as for control of biological
systems.1 The dihydroazulene/vinylheptafulvene molecuꢀ
lar photo/thermoswitch (1DHA/1VHF; Scheme 1) can unꢀ
dergo a 10ꢀπ electrocyclic reaction where significant
structural changes occur upon the ring opening/closure
reactions.1f,2 This system has also been recently identiꢀ
fied as a possible material on which to base molecular
solarꢀthermal (MOST) energy storage technologies.3
Currently, the major challenges for the exploitation of
solar energy are its efficient capture, storage, and onꢀ
demand release. MOST systems take advantage of the
photoactive nature of certain chemical compounds, and
although real applications have not yet matured, fundaꢀ
mental structureꢀproperty relationships have been estabꢀ
lished for various systems as well as promising proofꢀofꢀ
concept demonstration setꢀups.3,4 Solar energy absorpꢀ
tion by such materials is accompanied by a photochemiꢀ
cal isomerization forming a highꢀenergy photoisomer,
hence converting incident radiant energy into chemical
energy (Figure 1). In the case of electrocylic reactions,
this energy is stored in the formation of bonds between
atoms, which can be subsequently broken to release the
energy as heat. The DHAꢀVHF couple is particularly
attractive for such “closed cycle” thermal fuels (involving
Scheme 1. Photo/thermal switching of the parent
1DHA/1VHF and dimeric 2DHAꢀDHA/2VHFꢀVHF systems. The
commonly used numbering scheme is indicated.
NC
CN
8
CN
7
hv
NC
CN
CN
1
8a
3a
6
2
ꢀ
3
5
4
1DHA
sꢀcis 1VHF
sꢀtrans 1VHF
CN
CN
NC
CN NC
CN
hv
CN NC
ꢀ
2DHAꢀDHA
2VHFꢀVHF
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 2 of 13
two VHF units in a macrocycle presents an alternative
1
2
3
4
5
6
7
8
approach by providing energy discharge on two timeꢀ
scales via an initially fast VHF ring closure followed by a
slower one. If designed properly, the overall heat release
could be larger than for a nonꢀtethered system although
the second, slow discharging may still be of lower energy
than the first. One important property of the DHAꢀVHF
system remains the preꢀequilibrium between sꢀcis and sꢀ
trans conformers of VHF. While the sꢀtrans conformer is
usually more thermodynamically stable, only the sꢀcis
conformer has the geometry needed for ring closure.
Tuning the position of this equilibrium towards the sꢀ
trans conformer, possibly via constraints imposed by
macrocyclic structures, may be a convenient tool for
enhancing the lifetime of the VHF in spite of a smaller
activation energy for the sꢀcis conformer’s ring closure
reaction.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Here we set out to conduct a more comprehensive inꢀ
vestigation of the influence of macrocyclic ring strain on
the thermodynamic and kinetic properties of the
DHA/VHF system. This was undertaken through the
synthesis and characterization of three macrocyclic sysꢀ
tems (3DHAꢀDHA, 4DHAꢀDHA, and 5DHAꢀDHA; Figure 2), varyꢀ
ing only in the length of the bridging unit connecting the
two ends of the dimeric 2DHAꢀDHA system shown in
Scheme 1. Compound 2DHAꢀDHA can be easily accessed8
and was recently shown to be a convenient precursor for
Figure 1. Simplified potential energy surface for acyclic and
cyclic, dimeric systems converting between DHAꢀDHA
(blue) and VHFꢀVHF (red) isomers.
4DHAꢀDHA.
3b Here we show how this route can be generalꢀ
ized to allow the synthesis of macrocycles of various
sizes.
Two major challenges for MOST systems to be effective
are achieving a sufficiently high energy storage capacity
and control over the energy discharge. Taking various
considerations into account, the maximum energy densiꢀ
ty of a solarꢀthermal battery based on the isomerization
of an organic system has been estimated to be limited to
ca. 1 MJ kgꢀ1 (depending, however, on the absorption
maximum of the photoswitch).4c This energy storage
capacity corresponds to that of the unsubstituted norꢀ
bornadieneꢀquadricyclane (NB/QC) system.4a
O
O
O
O
O
O
O
O
5DHAꢀDHA
4DHAꢀDHA
CN
CN NC
NC
CN
CN NC
NC
Although the efficiency of the photoꢀinduced ring openꢀ
ing has been found to be generally quite high for many
DHAs (photoisomerization of 1DHA occurs with a quanꢀ
tum yield of 55%2b), calculations3a estimate that the enꢀ
ergy density of the parent 1DHA/1VHF system lies at ca.
O
O
O
O
0.11 MJ kgꢀ1. In addition, the halfꢀlife (1VHF t = 218 min
½
3DHAꢀDHA
in MeCN5) of the thermal discharge is too low with reꢀ
gard to practical applications, at least in a liquid medium
(e.g. heat transfer fluids of concentrated solar power
plants).
CN
NC
CN NC
Computational work by Durgun and Grossman4d on azoꢀ
benzenes6 and recent experimental results from our
lab3b on DHAs have suggested that the incorporation of
photoactive moieties into strained macrocyclic systems
could be a viable strategy towards increasing the Gibbs
free energy between the lightꢀharvesting DHA and metaꢀ
stable VHF states as well as providing a method for
controlling the energy release (Figure 1). Established
kinetic/thermodynamic postulates (i.e. BellꢀEvansꢀ
Polanyi principle7), however, imply that increasing the
relative energy difference between the two states will
also undesirably result in a lower activation energy for
the exothermic back reaction (VHF ꢀ DHA). Nevertheꢀ
less, our recent report3b has indicated that incorporating
Figure 2. Target macrocycles 3DHAꢀDHA, 4DHAꢀDHA, and
5DHAꢀDHA
.
Results and Discussion
Synthesis and Characterization. The syntheses of the
three macrocyclic DHA systems are shown in Scheme 2.
In each case, we start from the known diol 6,3b which is
readily made via two Suzuki couplings on the dibromoꢀ
substituted derivative of 2DHAꢀDHA. As previously deꢀ
scribed, diol 6 is difficult to obtain pure, as it is unstable
to column chromatography.3b Each of the three macroꢀ
cyclic DHA systems (3DHAꢀDHA, 4DHAꢀDHA, and 5DHAꢀDHA
)
ACS Paragon Plus Environment

Page 3 of 13
The Journal of Organic Chemistry
mers of all three cycles 3DHAꢀDHA (1732; 1160 cmꢀ1),
exists as a pair of diastereoisomers (a/b), the synthesis
4DHAꢀDHA (1732; 1163 cmꢀ1), and 5DHAꢀDHA (1730; 1162
1
2
3
4
5
6
7
8
and purification of which could be accomplished in a
similar way to that recently reported. The previously
reported synthesis conditions involve slow, yet simultaꢀ
neous addition of two solutions (diol 6 and a diacyl chloꢀ
ride) to a third solution of pyridine in CH2Cl2, which was
then stirred overnight at rt.3b Minor changes in the synꢀ
thesis of the key macrocyclization step could be
achieved using N,Nꢀdimethylaminopyridine (DMAP) to
promote the diester formation, reducing the reaction time
from overnight to only a few hours. The macrocycles
were separated from any byproducts (presumably includꢀ
ing oligomers) by column chromatography on silica gel
using 1% acetic acid in toluene as eluent. Minor impuriꢀ
ties, which seemed to be macrocyclic azulene byprodꢀ
ucts resulting from elimination of HCN, were subseꢀ
quently removed by size exclusion chromatography. An
increase in flexibility with increasing ring size (3DHAꢀDHA
to 4DHAꢀDHA to 5DHAꢀDHA) seems to be reflected in the
yields of the macrocyclization reactions; 13% (3ab), 17%
(4ab), and 29% (5ab). It is also interesting to note that
the melting points9 of the small macrocycles 3a and 3b
are higher than those of the large macrocycles 5a and
5b (3aDHAꢀDHA: 150ꢀ155 °C; 4aDHAꢀDHA: 138ꢀ150 °C;
5aDHAꢀDHA: 131ꢀ136 °C; 3bDHAꢀDHA: 156ꢀ163 °C; 4bDHAꢀ
DHA: 160ꢀ166 °C; 5bDHAꢀDHA: 128ꢀ134 °C), but for all
compounds quite broad ranges are obtained.
cmꢀ1).
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3. Molecular structures of macrocycles 3bDHAꢀDHA
(CCDC 1570242), 4aDHAꢀDHA (CCDC 1450505), and 5aDHAꢀ
DHA (CCDC 1570243). Hydrogen atoms were omitted in the
optimizations of 3bDHAꢀDHA and 5aDHAꢀDHA
soids are shown at 50% probability level. Single crystals
were grown from CHCl3/heptane (3bDHAꢀDHA or
. Thermal ellipꢀ
Scheme 2. Synthesis of macrocycles 3DHAꢀDHA, 4DHAꢀ
DHA, and 5DHAꢀDHA as pairs of diastereoisomers a/b.
)
CHCl3/EtOH (4aDHAꢀDHA, 5aDHAꢀDHA). Solvent molecules
were removed for clarity. Full asymmetric units are shown in
SI.
2DHAꢀDHA
O
O
1
The Hꢀ and 13CꢀNMR spectra of all macrocycles are
n
OH
HO
O
O
completely consistent with their structures. All protons in
the core DHA dimer seem slightly but consistently shifted
downfield with increasing macrocycle size for both diaꢀ
stereoisomers, δ 3DHAꢀDHA < 4DHAꢀDHA < 5DHAꢀDHA. The
same trend is observed for the 13CꢀNMR resonances of
the carbonyl carbon atoms. The two diastereoisomers
a/b for each macrocycle can also be distinguished by 1Hꢀ
and 13CꢀNMR spectroscopies. All 1H resonances of
3bDHAꢀDHA, for example, are systematically shifted upfield
compared to 3aDHAꢀDHA. This trend is not as encompassꢀ
ing for the other two macrocycles although most resoꢀ
nances of b diastereoisomers are upfield shifted relative
those of the a’s.
O
O
Diastereoisomers
a: RR / SS
(CH2)n
b: RS (meso)
Cl
Cl
CN
CN
CN NC
DMAP
CH2Cl2
4h, rt
CN
CN NC
NC
*
*
n = 0: 3DHAꢀDHA 13% (a+b)
n = 1: 4DHAꢀDHA 17% (a+b)
n = 2: 5DHAꢀDHA 29% (a+b)
6
Although single crystals could be grown from solutions of
the macrocycles presented here, only few samples were
suitable for Xꢀray diffraction (Figure 3). Furthermore,
even these exhibited conformational disorder and altꢀ
hough the dimeric DHA moieties are well resolved, the
overall structures could not be sufficiently optimized due
to considerable disorder in the bridging alkyl chain and
encapsulated solvent molecules. The structures do still
provide evidence for increased ring strain with decreasꢀ
ing macrocycle size as can be observed in a rather
“loose” conformation of 5DHAꢀDHA compared to the more
The UVꢀvis absorption spectra of macrocycles 3aDHAꢀ
DHA, 4aDHAꢀDHA, and 5aDHAꢀDHA show no substantial difꢀ
ferences between the wavelength or intensity of their
lowest energy absorptions (Figure 4). All macrocycles
exhibit characteristic absorptions assigned to the DHAꢀ
DHA moieties at ca. λmax = 356 nm exhibiting molar abꢀ
sorptivities around ε = 28000 Mꢀ1cmꢀ1. As expected, theꢀ
se are approximately twice as large as that of 1DHA, sugꢀ
gesting that the chromophores of the DHAꢀDHA macroꢀ
cycles can be considered as a sum of two individual
1DHA units. Furthermore, as all DHAꢀDHAs present simiꢀ
lar molar absorptivities, it can be concluded that these
exhibit no significant conformational distortion imposed
by cyclic strain, compared to 1DHA, even in the smallest
system (3), in line with the molecular structures achieved
3b
“taut” 4DHAꢀDHA and 3DHAꢀDHA, which clearly exhibits a
strained, “slingꢀshot” conformation in its structure.
Differences in ring strain might also be expected to disꢀ
tort/shift the interatomic stretching/bending vibrations of
these cycles; however, IR (ATR) spectroscopy shows no
effective differences in the intensities or shifts of the
absorptions. For comparison, carbonyl (C=O and CꢀO)
stretches overlap nearly perfectly in both diastereoisoꢀ
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 4 of 13
by Xꢀray crystallography. The absorptions are, however,
for 3ꢀ5aDHAꢀDHA, the parent 1DHA was also photoisomerꢀ
ized before each macrocyclic system to provide a referꢀ
ence, ref. The relative ring opening efficiencies can be
reasonably approximated by the τ2 τ1 ratio as well as
τ2 ref and τ1 ref. These data are summarized in Table 1.
1
2
3
4
5
6
7
slightly broader with increased ring size, observable in
both diastereoisomers of 4DHAꢀDHA and 5DHAꢀDHA comꢀ
pared to those of 3DHAꢀDHA. Such broadening could be
the result of greater flexibility and degrees of freedom in
the larger macrocycles.
τ
/
/
τ
/τ
Scheme 3. Interconversions between the three difꢀ
ferent isomers of the macrocyclic systems.
50000
3aDHAꢀDHA
4aDHAꢀDHA
8
9
40000
O
n
O
5aDHAꢀDHA
O
n
O
ε
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
30000
20000
10000
0
O
O
O
O
light
τ1
heat
NC
CN
CN NC
CN
2
t1/2
CN
CN
DHAꢀDHA
DHAꢀVHF
250
300
350
400
450
500
NC
Wavelength λ (nm)
O
n
O
light
τ2
heat
1
Figure 4. Measured UVꢀVis absorption spectra of macrocyꢀ
cles 3aDHAꢀDHA, 4aDHAꢀDHA, and 5aDHAꢀDHA in MeCN. Those
of 3ꢀ5bDHAꢀDHA are shown in the SI (Fig S21).
t1/2
O
O
NC
CN
Upon photoisomerization of dimer 2DHAꢀDHA to 2VHFꢀVHF, it
is observed that the molar absorptivity of 2VHFꢀVHF is
roughly double that of monomeric 1VHF.8 Considering the
VHFꢀVHF states of the strained macrocyclic structures,
3VHFꢀVHF, 4VHFꢀVHF, and 5VHFꢀVHF, however, it is observed
(vide infra) that the molar absorptivities of the second
VHF do not reach nearly the same intensity as would be
expected of two independent and nonꢀdisturbed VHF
chromophores. This indicates that ring strain in these
species might be playing a significant role in distorting
the structures. Thus far, characterization of the three
macrocyclic systems (3DHAꢀDHA, 4DHAꢀDHA, and 5DHAꢀDHA
a/b) has indicated that decreasing the length of the
diester linker does introduce some ring strain in the
DHAꢀDHA states, although a greater effect is hypotheꢀ
sized in the more conformationally elongated VHFꢀVHF
states.
CN
NC
VHFꢀVHF
2.2
2.0
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
5VHFꢀVHF
Ring Opening (τ2
5DHAꢀVHF
Ring Opening (τ1
)
5bDHAꢀDHA
)
300
400
500
600
700
800
Ring Opening Kinetics. All macrocycles were found to
undergo stepwise photoisomerization reactions from
DHAꢀDHA to DHAꢀVHF to VHFꢀVHF by irradiation at 356
nm (Scheme 3). This process could easily be followed by
UVꢀVis absorption spectroscopy by monitoring the
emergence of the characteristic VHF absorption at 494
nm (Figure 5). Due to the chiral nature of the DHA moieꢀ
ty, each of the three DHAꢀDHA macrocycles exists as a
pair of diastereoisomers (a and b), which are in fact
observed to exhibit significantly different ring opening
kinetics. Upon ring opening of the DHA moieties, howevꢀ
er, the chirality is lost resulting in only one pair of enantiꢀ
omers for the DHAꢀVHF state and none for the final
VHFꢀVHF state.
Wavelength λ / nm
2.0
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0
1000
2000
3000
4000
The kinetics of the photoisomerization reactions could be
modeled by a sum of two exponential functions characꢀ
terized by two time constants (τ1 and τ2), which, to an
approximation, relate to the two individual ringꢀopening
reactions (Scheme 3). To calibrate these time constants
Irradiation Time (s)
Figure 5. Top: Measured UVꢀVis absorption spectra
showing the photoisomerization of 5bDHAꢀDHA to 5DHAꢀVHF
ACS Paragon Plus Environment

Page 5 of 13
The Journal of Organic Chemistry
Structural distortion of the VHF chromophores away from
to 5VHFꢀVHF in MeCN by irradiation at 356 nm. Bottom:
Absorbance at = 494 nm plotted as a function of irradiꢀ
ation time and fitted by a biexponential function (red
curve): A(t) = c1 exp(ꢀt/ 1) + c2 exp(ꢀt/ 2) + A∞.
1
2
3
4
5
6
7
8
planarity results in poorer πꢀelectron delocalization and
thus significantly lower molar absorptivities with slightly
blueꢀshifted λmax values (Figure S22).10 The sꢀcis VHF
chromophore is also calculated to exhibit a lower molar
absorptivity, again likely a result of poorer πꢀconjugation
due to nonꢀplanarity. Compared to the more rigid DHAs,
the VHF chromophores seem to be more prone to conꢀ
formational distortion due to their greater flexibility.
Computational models of the HOMO and LUMO orbitals
of the DHA and sꢀtrans VHF isomers are presented in
the SI (Figure S27).
λ
τ
τ
Table 1. Ratios between time constants τref, τ1, and τ2
describing the photochemical ring opening reactions
of 1DHA (ref) and macrocycles 3DHAꢀDHA, 4DHAꢀDHA, and
5DHAꢀDHA. Values were obtained by fitting the time
evolution of the characteristic VHF absorbance by
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
the exponential function: A(t) = c1 exp(ꢀt/
τ
1) + c2 exp(ꢀ
t/τ2) + A∞ as in Figure 5.
Upon photochemical ring opening of the macrocyclic
DHAꢀDHAs, the DHAꢀVHF isomers are consistently
3a
4a
4b
5a
5b
observed to exhibit molar absorptivities of ca. ε356 nm
≈
19000 Mꢀ1 cmꢀ1 and ε498 nm ≈ 32000 Mꢀ1 cmꢀ1. By considꢀ
ering the two DHA and VHF chromophores independentꢀ
ly, it is observable that the absorption spectrum of DHAꢀ
VHF resembles that obtained by adding the absorptions
τ2
τ1
/
/
τ 1
39.2
0.96
37.8
59.6
0.75
44.7
34.4
1.3
49.4
0.73
36.1
33.1
τ ref
τ 2
/
τref
45.6
of 1DHA
( (ε472
ε354 nm = 16500 Mꢀ1 cmꢀ1) and sꢀtrans 1VHF
= 32000 Mꢀ1 cmꢀ1) as shown in Figure 7. This comꢀ
nm
The results reveal that the first ring opening exhibited by
all DHAꢀDHA macrocycles is comparable to that of the
1DHA/1VHF reference system and also that this first ring
parison serves to suggest that the core macrocyclic
DHAꢀVHF structure can be approximated by one strucꢀ
turally undisturbed 1DHA chromophore and a planar,
relatively undistorted sꢀtrans 1VHF chromophore.
opening is significantly faster than the second (τ1<<τ2).
5DHAꢀVHF
Structural/Spectral Correlations. Based on a recently
published computational and experimental investigation
of conformationally distorted DHA/VHF systems, it can
be reasonably concluded that a strong correlation exists
between the structure of the DHA/VHF chromophores
and their respective UVꢀVis absorption spectra.10 Strucꢀ
45000
1DHA + sꢀtrans 1VHF
40000
35000
30000
ε
25000
20000
15000
10000
5000
0
turally, the characteristic absorptions of 1DHA
nm, ε354 nm = 16500 Mꢀ1 cmꢀ1) and sꢀtrans 1VHF
472 nm,
472 nm = 32000 Mꢀ1 cmꢀ1) can be assigned to the
(λmax = 354
(λmax
=
ε
HOMO/LUMO transition corresponding to the relatively
planar and undisturbed chromophores highlighted in
Figure 6 (see SI for details).
300
400
500
600
700
Wavelength λ (nm)
40000
35000
sꢀtrans 1VHF
Figure 7. Measured molar absorptivities of 5DHAꢀVHF and a
sum of 1DHA + sꢀtrans 1VHF as a function of wavelength in
MeCN. Also shown is a representation of a possible 5DHAꢀ
VHF structure giving rise to the 5DHAꢀVHF spectrum.
30000
25000
20000
1DHA
15000
10000
5000
0
Notably, the second ring opening reaction from DHAꢀ
VHF to VHFꢀVHF results in only a small VHF absorptivity
increase as well as a slight blueshift of the λmax to 492
nm (ε
≈ 37000 Mꢀ1 cmꢀ1). Concurrently, the DHA absorpꢀ
tion at 360 nm is also observed to decrease only slightly
(Figure 8, top). Structurally, ring opening of the second
DHA moiety to VHF is hypothesized to result in inꢀ
creased cyclic strain which is alleviated through conforꢀ
mational distortion in one of the two VHF units. The reꢀ
sulting VHFꢀVHF structure may thus be approximated by
one VHF maintaining an sꢀtrans conformation and anꢀ
other exhibiting a structurally distorted and likely sꢀcis
VHF conformation (vide infra).
300
400
500
600
700
800
Wavelength λ (nm)
Figure 6. Measured UVꢀVis spectra of 1DHA (racemic mixꢀ
ture) and 1VHF (sꢀcis/sꢀtrans equilibrium mixture) in MeCN at
25 °C. The chromophores giving rise to the lowest energy
absorptions are highlighted in blue for 1DHA and red for sꢀ
trans 1VHF (which is the dominant conformer present).
Considering sꢀtrans 7VHF as a model system for a disꢀ
torted VHF moiety, it becomes reasonable that only a
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 6 of 13
isomerization from 5VHFꢀVHF to 5DHAꢀVHF to 5DHAꢀDHA a/b at
small absorptivity increase is observed for the characterꢀ
1
2
3
4
5
6
7
8
istic VHFꢀVHF absorption of the macrocyclic structures
(Figure 8). Furthermore, calculations predict a higher
molar absorptivity for sꢀtrans 7VHF compared to sꢀtrans
40 °C in MeCN is exemplified in Figure 9. Although theoꢀ
retically, the isomerizations are complicated by several
processes (vide infra), the decays monitored here could
be quite accurately modeled by a sum of two exponential
functions, assigned to the two thermal conversions
shown in Scheme 3. Although this is an approximation,
the assignment of each exponential to an individual VHF
decay is reasonable due to the large differences in halfꢀ
lives.
1VHF at ca. 310 nm, which corresponds to
a
HOMO/LUMO(+1) transition. Both of these orbitals are in
fact located exclusively on the 7ꢀmembered ring and this
absorption could in fact coincide with the decreasing
DHA absorption during the macrocylic DHAꢀVHF to VHFꢀ
VHF isomerization. Thus, these two simultaneous proꢀ
cesses could provide a rationale for why only a small
absorptivity decrease is observed experimentally in the
350 nm region for this ring opening (Figure 8). Additionꢀ
ally, the spectral sum of an undisturbed sꢀtrans 1VHF
chromophore and of the structurally distorted 7VHF bears
striking similarity to the spectrum attributed to the VHFꢀ
VHF state, lending further support to this hypothesis.
9
2.5
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
5VHFꢀVHF
2.0
1st Ring Closure
5bDHAꢀDHA
5DHAꢀVHF
1.5
1.0
0.5
0.0
2nd Ring Closure
5aDHAꢀDHA
5DHAꢀDHA
a/b after
45000
5DHAꢀVHF
40000
5VHFꢀVHF
one cycle
35000
30000
300
400
500
600
700
800
ε
25000
20000
15000
10000
5000
0
Wavelength λ / nm
2.5
2.0
1.5
1.0
0.5
300
400
500
600
700
800
Wavelength λ (nm)
sꢀtrans 1VHF
sꢀtrans 7VHF
45000
40000
35000
30000
25000
20000
15000
10000
5000
40 °C
MeCN
Sum of sꢀtrans 1VHF
and sꢀtrans 7VHF
0
500
1000
1500
2000
Time (min)
ε
Figure 9. Top: Measured UVꢀVis absorption spectra showꢀ
ing thermal conversion from 5VHFꢀVHF to 5DHAꢀVHF to 5DHAꢀ
a/b at 40 °C and the initial 5bDHAꢀDHA spectrum in
DHA
MeCN. Measurements were conducted in a sealed ampule
without baseline correction. Bottom: Absorbance at = 494
λ
nm plotted as a function of time and modeled by a sum of
two exponential decay functions (red curve) using the exꢀ
pression A(t) = c1 exp(ꢀkfastt) + c2 exp(ꢀkslowt) + A∞.
0
300
400
500
600
700
Wavelength λ (nm)
Figure 8. Top: Measured molar absorptivities of macrocycle
5 as a function of wavelength in MeCN. Bottom: Calculated
molar absorptivities of sꢀtrans 1VHF, sꢀtrans 7VHF10, and the
sum of the two in MeCN (CAMꢀB3LYP/6ꢀ311+G(d,p) using
the IEFPCM solvent model).
Estimates of activation energies (Ea) and preꢀexponential
factors (A) for the two processes (VHFꢀVHF to DHAꢀVHF
to DHAꢀDHA) were achieved using Arrhenius plots. In
the same manner, Eyring plots (Figure 10) were used to
estimate enthalpies (ꢀH‡ ) and entropies (ꢀS‡ ) of activaꢀ
tion, from which Gibbs free energies of activation (ꢀG‡ )
were calculated. It is notable that although the decay
model achieves a great fit for the least strained macroꢀ
Ring Closure Kinetics. Thermal isomerizations were
conducted for macrocycles 3VHFꢀVHF, 4VHFꢀVHF, and 5VHFꢀ
from VHFꢀVHF to DHAꢀVHF to DHAꢀDHA at 25, 40,
cycle (5VHFꢀVHF to 5DHAꢀVHF to 5DHAꢀDHA), the fit somewhat
VHF
50, 60, 70 and 80 °C in MeCN. The kinetics of all therꢀ
mal conversions were monitored by UVꢀVis spectroscoꢀ
py and observed to proceed in a stepwise manner via an
initially fast decay followed by a slower one. The thermal
worsens with increased strain. In fact, the slowest decay,
which was recorded for 3DHAꢀVHF to 3DHAꢀDHA over 45
days at rt, exhibited a large enough deviation from the
model that the initial and more sensitive decay from 3VHFꢀ
ACS Paragon Plus Environment

Page 7 of 13
The Journal of Organic Chemistry
proceeds to yield a pair of diastereoisomers which
to 3DHAꢀVHF could not be reliably used in the calculaꢀ
VHF
1
2
should, in principle, be theoretically modeled as two
processes. Additionally, the sꢀcis / sꢀtrans VHF equilibriꢀ
um likely also plays a role, further slowing the thermal
ring closure if the molecules cannot easily get from the
usually more stable sꢀtrans to the reactive sꢀcis conforꢀ
mation. In fact, this process can be used to explain the
large difference between the first and second VHFꢀtoꢀ
DHA ring closure rates. Structurally, the VHFꢀVHF states
are thus hypothesized to present both a distorted VHF
(close to sꢀcis based on computational results, vide infra)
and an sꢀtrans VHF moiety.
tion of kinetics parameters. This data point was thus not
included in the Arrhenius or Eyring plots.
3
4
5
6
7
8
9
5VHFꢀVHF kfast
ꢀ10
4VHFꢀVHF kfast
3VHFꢀVHF kfast
ꢀ12
ꢀ14
ꢀ16
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
5DHAꢀVHF kslow
The isomerization from VHF to DHA may in principle
proceed via two sites of ring closure, accessible by rotaꢀ
tion about the exocyclic C3ꢀC3a fulvene bond of VHF
achievable via VHF E/Z isomerization. Consequently, the
site of the linker bridging the two sevenꢀmembered rings
can isomerize from the 7ꢀ to the 6ꢀposition in one or both
of the DHA moieties of the DHAꢀDHA state. Previous
reports on nonꢀcyclic structures have indeed identified
partial 7/6ꢀisomerization to occur as a result of a lightꢀ
heat cycle (ringꢀopening – ringꢀclosure) yielding mixtures
of two DHA isomers – evidenced by a significant redshift
in the DHA absorption maxima.1f The reaction rate modꢀ
el proposed here is only an approximation for two genꢀ
eral reaction pathways, which we tentatively assign to an
initially fast VHFꢀVHF to DHAꢀVHF ring closure followed
by a slower DHAꢀVHF to DHAꢀDHA ring closure. Altꢀ
hough the thermal reaction was not carried out to full
completion in the experiment shown in Figure 9, it is
clear that differences in the spectra are minimal. This
reinforces our approximation that the thermal isomerizaꢀ
tion from VHFꢀVHF to DHAꢀDHA primarily proceeds to
yield a diastereoisomeric mixture of a/b without any
significant byproducts, which is also supported by TLC
analysis.
ꢀ18
4DHAꢀVHF kslow
3DHAꢀVHF kslow
ꢀ20
0.0028
0.0030
0.0032
0.0034
K / T
Figure 10. Eyring plots for the fast (VHFꢀVHF to DHAꢀVHF)
and slow (DHAꢀVHF to DHAꢀDHA) thermal isomerization
reactions of macrocycles 3VHFꢀVHF, 4VHFꢀVHF, and 5VHFꢀVHF
.
Thermodynamic parameters (Ea, A, ꢀH‡, ꢀS‡, ꢀG‡) in
addition to the rate constants (k) and halfꢀlives (t ) at
½
room temperature for both the fast and slow thermal
decay processes of the three macrocycles are summaꢀ
rized in Table 2. Comparing the rates of the two thermal
ring closures as a function of macrocycle size and ring
strain reveals that the ring closures systematically proꢀ
ceed slower (higher ꢀG‡’s) with decreasing macrocycle
size: 5VHFꢀVHF > 4VHFꢀVHF > 3VHFꢀVHF. Contrary to this, the
enthalpies of activation (ꢀH‡) for both the fast and slow
thermal ring closure processes were actually found to
decrease with decreasing macrocycle size. The observaꢀ
tions of decreasing ꢀH‡ but overall higher ꢀG‡ values
with decreasing macrocycle size can be reconciled by
considering the entropy of activation (ꢀS‡), which is
revealed to progressively become more negative. The
negative ꢀS‡ values also suggest that entropy decreases
upon forming the transition state, consistent with an
associative, ring closing mechanism, generating a more
ordered state. Secondly, progressively more negative
ꢀS‡ values with decreasing macrocycle size suggest that
more conformational reorganization is needed for the
smaller macrocycles to undergo ring closure than for the
larger. Due to the fact that the larger macrocycles exhibit
more degrees of freedom than the smaller, one VHFꢀtoꢀ
DHA ring closure reaction will eliminate a greater proporꢀ
tion of degrees of freedom in the small cycles compared
to the larger. For this reason, relatively greater reꢀ
organization, and thus larger ꢀS‡ values, are observed in
the smaller macrocycles, reflecting a higher energy reꢀ
quirement. Thus, the observation that the thermal ring
closures are disfavored with decreasing macrocycle size
can be rationalized by the fact that the increase in the
entropic ꢀS‡ contribution to the Gibbs free energy of
activation (ꢀG‡) outweighs the decreasing ꢀH‡. The
overall result is a higher activation barrier (ꢀG‡) in the
smaller macrocycles. The thermal ringꢀclosing isomerizaꢀ
tions are, however, theoretically complicated by several
other possible processes. Firstly, it should be noted that
the second ring closure from DHAꢀVHF to DHAꢀDHA
Calculations – Energy Landscape. For a complete
picture of the potential energy surface of the macrocyclic
systems, a computational investigation was undertaken
with the goal of providing estimates for macrocyclic geꢀ
ometries and relative energies between the DHAꢀDHA,
DHAꢀVHF and VHFꢀVHF states of systems 3, 4, and 5. A
systematic conformer search was first carried out using a
combination of the ConfGen program from MSTor11a and
Gaussian 0911b to systematically generate a library of
macrocyclic conformers.
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 8 of 13
1
2
3
4
5
6
7
8
ꢀH‡
ꢀS‡
ꢀG‡
298 K
Ea
A
k298 K
t
½
[kJ molꢀ1] [108 minꢀ1]
[10ꢀ5 minꢀ1]
[kJ molꢀ1]
59.2
[J Kꢀ1 molꢀ1 ]
ꢀ121
[kJ molꢀ1]
95.3
[min]
98
705
61.9
77.0
63.7
86.4
67.1
92.7
5.2
3
3
4
4
5
5
VHFꢀVHF ꢀ 3DHAꢀVHF
DHAꢀVHF ꢀ 3DHAꢀDHA
VHFꢀVHF ꢀ 4DHAꢀVHF
DHAꢀVHF ꢀ 4DHAꢀDHA
VHFꢀVHF ꢀ 5DHAꢀVHF
DHAꢀVHF ꢀ 5DHAꢀDHA
5.7
12100
77
23
74.3
61.0
83.7
64.4
89.5
ꢀ109
ꢀ110
ꢀ72.8
ꢀ97.5
ꢀ49.5
106.7
93.9
905
19
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
12.7
1180
17.8
5460
59
1730
99
105.4
93.5
3900
34200
104.2
Table 2. Experimentally determined thermodynamic parameters for the first (VHFꢀVHF to DHAꢀVHF) and second
(DHAꢀVHF to DHAꢀDHA) thermal isomerizations of macrocycles 3VHFꢀVHF 4VHFꢀVHF and 5VHFꢀVHF
,
,
.
Due to the cyclic structure of the macrocycles, bonds
could not be freely rotated and were therefore broken to
achieve rotation. Twoꢀ and threeꢀfold rotations about
single bonds were allowed according to Figure S23 (see
SI). The original connectivity was then reꢀimposed in
Gaussian 09 and a forceꢀfield (Dreiding) optimization
was run on each conformer to generate more than 1.3
million macrocyclic structures. Conformers presenting
interatomic distances of less than 0.5 Å were automatiꢀ
cally discarded by the MSTor program. For all macrocyꢀ
cles 3, 4, and 5, the 100 lowest energy conformers of
each DHA/VHF subunit orientation were subjected to
PM3 geometry optimizations and frequency calculations.
This corresponds to 600 DHAꢀDHA (300 a / 300 b), 800
DHAꢀVHF, and 1000 VHFꢀVHF conformers. The SI conꢀ
tains more details for the conformer selection proceꢀ
dures. The number of conformers used to proceed with
is representative of the various DHA and VHF conforꢀ
mations that are accessible by the macrocycles, not to
conformational variations in the alkyl bridge. Thus, fewer
“characteristic” DHAꢀDHA conformers are required to
achieve a representative sample of the conformational
energy landscape as this state is much more rigid than
the systems including VHF moieties. Notably, PM3 optiꢀ
mization results indicate that the relative energy differꢀ
ences between the lowest energy conformers of the
DHAꢀDHA, DHAꢀVHF, and VHFꢀVHF macrocycles are
very precise representations of the energy differences
considering a distribution (90% Boltzmann weighted
average) of the lowest energy conformers of each state
(See SI for details). Additionally, PM3ꢀcalculated relative
energies (ΔG) also predict a correct energetic stability
trend between the macrocycles, that is, ꢀG DHAꢀDHA <
DHAꢀVHF < VHFꢀVHF.
gies of other “characteristic” conformers are provided in
SI (Table S8).
The results obtained from DFT geometry optimizations
seem to provide reasonably accurate structural repreꢀ
sentations for the individual DHA/VHF chromophoric
units (Figure 11). This is judged on the basis of experiꢀ
mental UVꢀVis absorption spectra, which correlate well
with the conformations of these chromophores. For all
macrocyclic systems, the DHAꢀDHA state presents two
relatively undistorted core DHA chromophores. Similarly,
all lowestꢀenergy DHAꢀVHF systems present the hyꢀ
pothesized undistorted DHA moiety as well as the planar
and undistorted sꢀtrans VHF moiety. Lastly and again
according to hypothesis, all lowestꢀenergy macrocyclic
VHFꢀVHF conformers exhibit a planar, sꢀtrans VHF as
well as a partially outꢀofꢀplane, sꢀcis VHF moiety. These
observations not only explain the characteristic UVꢀVis
absorption spectra assigned to these states (sꢀtrans VHF
and distorted VHF of VHFꢀVHF) but also the initially fast
VHFꢀVHF to DHAꢀVHF thermal isomerization (distorted
VHF is close to an sꢀcis conformation) and the subseꢀ
quently slow DHAꢀVHF to DHAꢀDHA reaction (from the
sꢀtrans VHF).
The calculated lowestꢀenergy conformers for the macroꢀ
cyclic DHAꢀVHF and VHFꢀVHF systems are all found to
exhibit “ringꢀcollapsed” structures to various degrees
(characterized by reduced cavity space). This is a reaꢀ
sonable consequence of simulating structural optimizaꢀ
tions in vacuum without explicit solvent molecules. As
the rigid DHAꢀDHA macrocycles maintain an open strucꢀ
ture, a relative energy comparison between these and
the ringꢀcollapsed DHAꢀVHF or VHFꢀVHF macrocycles is
unfounded. In fact, the relative Gibbs free energies (ꢀG)
between the DHAꢀDHA and DHAꢀVHF macrocycles
seem unreasonable, as the data suggest a more stable
DHAꢀVHF than DHAꢀDHA for 3 and 4, contradictory to
experimental observations, which reveal DHAꢀDHA to be
To obtain more accurate Gibbs free energies (ꢀG), seꢀ
lected PM3ꢀoptimized structures were subjected to M06ꢀ
2X/6ꢀ31+G(d) geometry optimizations and frequency
calculations in vacuum. The lowestꢀenergy conformers
and relative energies are presented in Figure 11. Enerꢀ
o
thermodynamically most favorable at 25 C (as >75%
DHAꢀDHA seemed to have formed from DHAꢀVHF when
the experiment was stopped, and the ultraslow converꢀ
ACS Paragon Plus Environment

Page 9 of 13
The Journal of Organic Chemistry
sion was clearly not complete at this stage). Solvation
investigations, not only for advancing MOST systems but
in general for development of smart functional materials.
1
would likely have a larger stabilizing influence on the
DHAꢀDHA isomer than on the more flexible DHAꢀVHF
and VHFꢀVHF isomers for which favorable intramolecuꢀ
lar van der Waals interactions are already present in the
gas phase structures. Thus, in solution, the DHAꢀVHF
and VHFꢀVHF states will likely be more energetic relative
to the DHAꢀDHA state than in the gas phase. Optimizaꢀ
tions involving solvent modeling or guest molecules
occupying the macrocyclic cavities perhaps using
QM/MM methods12 might provide more accurate estimaꢀ
tions of the relative energy differences. Nevertheless,
considering the lowestꢀenergy DHAꢀDHA and VHFꢀVHF
conformers in the gas phase (Figure 11), we see that the
smallest, most strained macrocycle 3 clearly exhibits the
most energetic VHFꢀVHF state: 60 kJ molꢀ1 (3VHFꢀVHF).
This corresponds to 30 kJ per mole VHF. This value is
similar to the parent system 1 (28 kJ molꢀ1), but since the
gas phase VHFꢀVHF structure is more collapsed than
the DHAꢀDHA structure, it is expected that solvation
would in actual fact result in an increased energy storage
of the macrocycle. Notably, the largest macrocycle 5 has
a more energetic VHFꢀVHF state than 4 in the gas
phase, so a general correlation between ring size and
energy density is more complicated.
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Conclusion
The switching properties of two metaꢀphenyleneꢀbridged
DHA units can be finely tuned by incorporation into macꢀ
rocyclic structures of various sizes. Thus, we find that
cyclization causes stepꢀwise, photoinduced ring opening
reactions to the VHFꢀVHF isomer and subsequent stepꢀ
wise thermal ring closure reactions. Particularly interestꢀ
ing, we were able to tune the halfꢀlife of the second VHF
ring closure reaction (DHAꢀVHF to DHAꢀDHA) in the
macrocycles from 65 h (5) to 91 h (4) to 202 h (3) at
room temperature by decreasing the macrocyclic ring
size. The hampering of the second VHF ring closure
reaction seems in particular to be the result of an unfaꢀ
vorable entropic term, increasing with increased ring
strain. By combining the current macrocyclic approach
with other structural modifications, it may be possible to
set the back reaction completely on standꢀby. For ultiꢀ
mate release of the energy when needed, we may emꢀ
ploy our recent discovery that the back reaction is proꢀ
moted by the action of Cu(I) ions.13 We are currently also
targeting other triggering mechanisms. Indeed, a fast
thermal discharge upon activation is needed in order for
a MOST system to deliver high power.
It is more difficult to find a correlation between ring size
and energy density, at least based on the gas phase
structures reported in this work. Yet, the smallest macroꢀ
cycle in the series has the largest energy difference
between DHAꢀDHA and VHFꢀVHF states. For increasing
the overall energy density of the system, lowꢀenergy,
ringꢀcollapsed conformers should ideally be avoided.
Collapses are likely avoided by solvation or by incorpoꢀ
rating the macrocycles in mechanically interlocked strucꢀ
tures, such as rotaxanes and catenanes.6c,14 Such stratꢀ
egies could reveal the further assets of photoswitchable
macrocycles and should certainly be the scope of future
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 10 of 13
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 11. Calculated (M06ꢀ2X/6ꢀ31+G(d)) lowestꢀenergy conformations of macrocycles 3, 4, and 5 and Gibbs free energies
(ꢀG/kJ molꢀ1; in brackets) relative to the lowestꢀenergy DHAꢀDHA conformers. Hydrogen atoms were removed for clarity. For
reference system 1DHA/sꢀtrans 1VHF, ꢀG = 28 kJ molꢀ1.
acquired on an ATR single reflection diamond instruꢀ
Experimental Section
ment. Samples were loaded directly or by evaporation
from a suitable solvent. IR absorptions are reported in
units of wavenumbers (cmꢀ1). Mass spectra were recordꢀ
ed by using a MALDIꢀFTꢀICR instrument equipped with a
7 T magnet (prior to the experiments, the instrument was
calibrated using sodium trifluoroacetate (NaTFA) cluster
ions). Thin layer chromatography (TLC) was conducted
using commercially available, precoated plates (silica 60)
with fluorescence indicator. With respect to the DHA
compounds, TLC was carried out in the absence of light;
a color change from yellow to red upon exposure to UV
light indicates a conversion to VHF. Size exclusion colꢀ
umn chromatography was conducted on SꢀX8 polystyꢀ
rene biobeads using CH2Cl2 as eluent. All melting points
are uncorrected.
General Methods. All lightꢀsensitive compounds, reacꢀ
tions and manipulations were shielded from light by eiꢀ
ther conducting the procedures in a dimly lit room or by
masking the glassware and equipment with aluminum
foil. Common solvents were dried, distilled, and deꢀ
gassed prior to use as follows: CH2Cl2 was dried using a
solvent purification system; CH2Cl2 was alternatively
dried over 3 Å molecular sieves. Commercially available
reagents including 4ꢀdimethylaminopyridine (DMAP),
suberoyl chloride, and dodecanedioyl dichloride were
purchased from commercial sources and used as reꢀ
ceived
without
further
purification.
2,2'ꢀ(1,3ꢀ
Phenylene)bis(7ꢀ(4ꢀ(hydroxymethyl)phenyl)azuleneꢀ
1,1(8aH)ꢀdicarbonitrile) (6) was synthesized with only
partial purification, according to literature procedure.3b
Purification of products carried out by flash column
chromatography was conducted using silica gel (60 Å
pore size; 43ꢀ60 ꢀm particle size). Nuclear magnetic
resonance (NMR) spectra were acquired on either a
500ꢀMHz instrument (1H at 500 MHz, 13C at 126 MHz)
equipped with a nonꢀinverse cryo probe or a pentaꢀ
probe/broadꢀband probe. Chemical shift values are proꢀ
General Macrocyclization Reaction Procedure. A
flameꢀdried, 3ꢀnecked, 250 mL rbf was equipped with
two 50 mL dropping funnels and charged with a solution
of 4ꢀdimethylaminopyridine (76 mg, 0.62 mmol) in dry
CH2Cl2 (40 mL) under N2 atmosphere. The dropping
funnels were loaded with two solutions of i) diꢀacid chloꢀ
ride (suberoyl chloride or dodecanedioyl dichloride) (0.19
mmol) in CH2Cl2 (40 mL) and ii) compound 6 (100 mg,
0.16 mmol). The contents of the dropping funnels were
then added dropwise to the reaction vessel over 30 min,
1
vided relative to internal solvent references for H and
13CꢀNMR spectra. Infrared spectroscopy (IR) data were
ACS Paragon Plus Environment

Page 11 of 13
The Journal of Organic Chemistry
and the reaction mixture was stirred at rt for another 3.5
5a: pale yellow solid (17 mg, 13%). Crystals suitable for
1
2
3
4
5
6
7
8
h in the absence of light. The reaction mixture was transꢀ
ferred to a separatory funnel and washed with 1 M HCl
(2 x 50 mL) followed by H2O (50 mL). The organic phase
was isolated, dried over MgSO4, filtered, and concenꢀ
trated in vacuo. The residue was purified by flash column
chromatography (SiO2, 1% AcOH/toluene) to isolate the
two diastereoisomers of the title compounds with minor
azulene impurities. The mixtures were further purified by
size exclusion column chromatography on SꢀX8 polystyꢀ
rene biobeads using CH2Cl2 as eluent. The products
Xꢀray diffraction were grown from layered CHCl3/EtOH
which was also used to determine stereochemistry. TLC
(40% EtOAc/toluene) Rf = 0.84. Mp: 120 °C darkens to
1
orange solid, 131 ꢀ 136 °C melts to orange liquid. H
NMR (500 MHz, CDCl3) δ 7.97 (t, J = 1.7 Hz, 1H), 7.80
(dd, J = 7.9, 1.7 Hz, 2H), 7.61 (t, J = 7.9 Hz, 1H), 7.38
(br d, JAB = 8.4 Hz, 4H), 7.34 (br d, JAB = 8.4 Hz, 4H),
6.94 (s, 2H), 6.87 (dd, J = 11.5, 5.8 Hz, 2H), 6.81 (d, J =
11.5 Hz, 2H), 6.43 (br d, J = 5.8 Hz, 2H), 5.91 (d, J = 4.8
Hz, 2H), 5.12 (d, JAB = 12.3 Hz, 2H), 5.08 (d, JAB = 12.3
Hz, 2H), 3.83 (dd, J = 4.8, 1.3 Hz, 2H), 2.30 – 2.26 (m,
4H), 1.60 – 1.55 (m, 4H), 1.27 – 1.18 (m, 12H) ppm. 13C
NMR (126 MHz, CDCl3) δ 173.6, 140.3, 139.9, 139.3,
139.2, 136.4, 133.0, 132.8, 132.0, 131.9, 130.4, 128.8,
128.0, 127.5, 124.5, 121.1, 117.1, 115.3, 112.9, 65.7,
51.0, 45.2, 34.7, 29.9, 29.7, 29.5, 25.1 ppm. IR: 3031w,
2926m, 2854m, 2254w, 2208w, 1730m, 1609w, 1513w,
1486w, 1455w, 1379w, 1350w, 1244m, 1216m cmꢀ1.
HRMS (MALDI+ FTꢀICR, Dithranol, m/z): found
841.37592 [M+H]+, calc. for [C56H49N4O4]+ 841.37483
[M+H]+.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
could
be
concentrated
from
a
mixture
of
CH2Cl2/heptanes to yield the title compounds as wellꢀ
behaved yellow solids.
3a: pale yellow solid (7 mg, 6%). Stereochemistry was
assigned as the complementary diastereoisomer to 3b.
TLC (40% EtOAc/toluene) Rf = 0.81. Mp: 120ꢀ140 °C
darkens to a red/orange solid, 150 ꢀ 155 °C melts to
1
red/orange liquid. H NMR (500 MHz, CDCl3) δ 7.91 (br
t, J = 1.7 Hz, 1H), 7.77 (dd, J = 7.9, 1.7 Hz, 2H), 7.60 (t,
J = 7.9 Hz, 1H), 7.34 (d, JAB = 8.3 Hz, 4H), 7.31 (d, JAB
=
8.3 Hz, 4H), 6.89 – 6.79 (m, 6H), 6.40 (br d, J = 5.5 Hz,
2H), 5.80 (d, J = 4.8 Hz, 2H), 5.11 (d, JAB = 12.2 Hz, 2H),
5.07 (d, JAB = 12.2 Hz, 2H), 3.80 (dd, J = 4.8, 1.2 Hz,
2H), 2.29 – 2.24 (m, 4H), 1.60 – 1.55 (m, 4H), 1.29 –
1.25 (m, 4H) ppm. 13C NMR (126 MHz, CDCl3) δ 173.4,
140.7, 140.0, 139.2, 138.6, 136.4, 133.4, 132.7, 132.4,
131.6, 130.4, 128.9, 127.9, 127.9, 126.0, 120.9, 117.5,
115.1, 113.0, 65.7, 51.0, 45.9, 34.6, 29.3, 25.0 ppm. IR:
3031w, 2928m, 2856m, 2252w, 1732s, 1655m, 1612w,
1567w, 1513m, 1487m, 1454m, 1413m, 1370m, 1292m,
1244m, 1213m cmꢀ1. HRMS (MALDI+ FTꢀICR, Dithranol,
m/z): found 785.31391 [M+H]+, calc. for [C52H41N4O4]+
785.31223 [M+H]+.
5b: pale yellow solid (21 mg, 16%). Crystals were grown
from layered CHCl3/heptanes and CHCl3/EtOH but neiꢀ
ther were suitable for Xꢀray diffraction. Stereochemistry
was assigned as the complementary diastereoisomer to
5a. TLC (40% EtOAc/toluene) Rf = 0.78. Mp: 120 °C
darkens to orange solid, 128 ꢀ 134 °C melts to orange
liquid. 1H NMR (500 MHz, CDCl3) δ 7.95 (t, J = 1.8 Hz,
1H), 7.83 (dd, J = 7.9, 1.8 Hz, 2H), 7.64 (t, J = 7.9 Hz,
1H), 7.37 (br d, JAB = 8.4 Hz, 4H), 7.33 (br d, JAB = 8.4
Hz, 4H), 6.88 (s, 2H), 6.84 (dd, J = 11.5, 5.6 Hz, 2H),
6.80 (d, J = 11.5 Hz, 2H), 6.38 (br d, J = 5.6 Hz, 2H),
5.94 (d, J = 4.8 Hz, 2H), 5.10 (s, 4H), 3.85 (dd, J = 4.8,
1.5 Hz, 2H), 2.31 – 2.27 (m, 4H), 1.60 – 1.53 (m, 4H),
1.26 – 1.17 (m, 12H) ppm. 13C NMR (126 MHz, CDCl3) δ
173.7, 140.8, 139.8, 139.4, 139.2, 136.5, 132.9, 132.6,
132.1, 132.1, 130.5, 128.7, 128.0, 127.6, 125.0, 120.9,
117.0, 115.0, 112.7, 65.7, 51.1, 45.4, 34.7, 29.8, 29.6,
29.5, 25.1 ppm. IR: 3061w, 3029w, 2926m, 2854m,
2255w, 2207w, 1730m, 1611w, 1513w, 1442w, 1417w,
1378w, 1351w, 1246m, 1217m cmꢀ1. HRMS (MALDI+
FTꢀICR, Dithranol, m/z): found 841.37727 [M+H]+, calc.
calc. for [C56H49N4O4]+ 841.37483 [M+H]+.
3b: pale yellow solid (8 mg, 7%). Crystals suitable for Xꢀ
ray diffraction were grown from layered CHCl3/heptane
which was also used to determine stereochemistry. TLC
(40% EtOAc/toluene) Rf = 0.77. Mp: 120ꢀ140 °C darkens
to a red/orange solid, 156 ꢀ 163 °C melts to red/orange
liquid. 1H NMR (500 MHz, CDCl3) δ 7.83 (dd, J = 7.9, 1.1
Hz, 2H), 7.68 (t, J = 7.9 Hz, 1H), 7.59 (br t, J = 1.1 Hz,
1H), 7.33 (d, JAB = 8.3 Hz, 4H), 7.30 (d, JAB = 8.3 Hz,
4H), 6.86 (dd, J = 11.4, 5.5 Hz, 2H), 6.81 (d, J = 11.4 Hz,
2H), 6.71 (s, 2H), 6.33 (br d, J = 5.5 Hz, 2H), 5.78 (d, J =
4.8 Hz, 2H), 5.08 (d, JAB = 12.1 Hz, 2H), 5.04 (d, JAB
=
AUTHOR INFORMATION
12.1 Hz, 2H), 3.79 (d, J = 4.8 Hz, 2H), 2.26 – 2.20 (m,
4H), 1.53 – 1.49 (m, 4H), 1.26 – 1.21 (m, 4H) ppm. 13C
NMR (126 MHz, CDCl3) δ 173.4, 141.6, 139.8, 138.9,
138.5, 136.4, 132.9, 132.8, 132.6, 131.6, 130.5, 129.1,
127.9, 127.5, 126.9, 120.5, 117.5, 115.1, 112.7, 65.8,
51.3, 45.8, 34.5, 29.1, 24.8 ppm. IR: 3058w, 3027w,
2929m, 2861m, 2250w, 1732s, 1654m, 1612w, 1513m,
1487m, 1455m, 1410m, 1368m, 1291m, 1260m, 1214m
cmꢀ1. HRMS (MALDI+ FTꢀICR, Dithranol, m/z): found
785.31391 [M+H]+, calc. for [C52H41N4O4]+ 785.31223
[M+H]+.
Corresponding Author
* Eꢀmail: mbn@chem.ku.dk
Author Contributions
The manuscript was written through contributions of all
authors.
Funding Sources
We thank University of Copenhagen for supporting this
work.
4a: pale yellow solid (10 mg, 8%). Characterization data
ASSOCIATED CONTENT
Supporting Information
are consistent with those previously reported.3b
4b: pale yellow solid (11 mg, 9%). Characterization data
are consistent with those previously reported.3b
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 12 of 13
13. (a) Cacciarini, M.; Vlasceanu, A.; Jevric, M.; Nielsen, M.
The Supporting Information is available free of charge on
the ACS Publications website at DOI: ###.
NMR/IR/UV−Vis spectra, switching studies, computational
data, Xꢀray crystallographic data (PDF)
Xꢀray crystallographic data for 3bDHAꢀDHA (CIF)
Xꢀray crystallographic data for 5aDHAꢀDHA (CIF)
B. Chem. Commun. 2017, 53, 5874ꢀ5877; (b) Wang, Z.; Udꢀ
mark, J.; Börjesson, K.; Rodrigues, R.; Abrahamsson, M.; Nielꢀ
sen, M. B.; MothꢀPoulsen, K. ChemSusChem 2017, 10, 3049ꢀ
3055.
14. Shilova, E. A.; Perevalov, V. P.; Suslov, V. V.; Moustrou,
C.; Tetrahedron Lett. 2008, 49, 3453ꢀ3457.
1
2
3
4
5
6
7
REFERENCES
1. (a) Tian, H.; Yang, S. J. Chem. Soc. Rev. 2004, 33, 85ꢀ97;
(b) Russew, M.ꢀM.; Hecht, S. Adv. Mater. 2010, 22, 3348ꢀ3360;
(c) Irie, M.; Fukaminato, T.; Matsuda, K.; Kobatake, S. Chem.
Rev. 2014, 114, 12174ꢀ12277; (d) Zhang, J.; Zhou, Q.; Tian, H.
Adv. Mater. 2012, 25, 378ꢀ399; (e) Szymánski, W.; Beierle, J.
M.; Kistemaker, H. A.; Velema, W. A.; Feringa, B. L. Chem.
Rev. 2013, 113, 6114ꢀ6178; (f) Broman, S. L.; Nielsen, M. B.
Phys. Chem. Chem. Phys. 2014, 16, 21172ꢀ21182.
2. (a) Daub, J.; Knöchel, T.; Mannschreck, A. Ang. Chem.
Int. Ed. Engl. 1984, 23, 960ꢀ961; (b) Görner, H.; Fischer, C.;
Gierisch, S.; Daub, J. J. Phys. Chem. 1993, 97, 4110ꢀ4117.
3. (a) Olsen, S. T.; Elm, J.; Storm, F. E.; Gejl, A. N.; Hansen,
A. S.; Hansen, M. H.; Nikolajsen, J. R.; Nielsen, M. B.;
Kjaergaard, H. G.; Mikkelsen, K. V. J. Phys. Chem. A 2015,
119, 896ꢀ904; (b) Vlasceanu, A.; Broman, S. L.; Hansen, A. S.;
Skov, A. B.; Cacciarini, M.; Kadziola, A.; Kjaergaard, H. G.;
Mikkelsen, K. V.; Nielsen, M. B. Chem. Eur. J. 2016, 22, 10796ꢀ
10800.
4. (a) Yoshida, Z.ꢀI. J. Photochem. 1985, 29, 27ꢀ40; (b)
Kanai, Y.; Srinivasan, V.; Meier, S. K.; Vollhardt, K. P. C.;
Grossman, J. C. Angew. Chem. Int. Ed. 2010, 49, 8926ꢀ8929;
(c) Kucharski, T. J.; Tian, Y.; Akbulatov, S.; Boulatov, R. Enerꢀ
gy Environ. Sci. 2011, 4, 4449ꢀ4472; (d) Durgun, E.; Grossman,
J. C. J. Phys. Chem. Lett. 2013, 4, 854ꢀ860; (e) Kucharski, T.
J.; Ferralis, N.; Kolpak, A. M.; Zheng, J. O.; Nocera, D. G.;
Grossman, J. C. Nature Chem. 2014, 6, 441ꢀ447; (f) Lenꢀ
nartson, A.; Roffey, A.; MothꢀPoulsen, K. Tetrahedron Lett.
2015, 56, 1457ꢀ1465; and references cited therein; g) Dreos,
A.; Börjesson, K.; Wang, Z.; Roffey, A.; Norwood, Z.; Kushnir,
D.; MothꢀPoulsen, K. Energy Environ. Sci. 2017, 10, 728ꢀ734.
5. Broman, S. L.; Brand, S. L.; Parker, C. R.; Petersen, M.
Å.; Tortzen, C. G.; Kadziola, A.; Kilså, K.; Nielsen, M. B.
ARKIVOC 2011, ix, 51ꢀ67.
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
6. For examples of macrocycles incorporating azobenzenes,
see: (a) Norikane, Y.; Tamaoki, N. Org. Lett. 2004, 6, 2595ꢀ
2598; (b) Bassotti, E.; Carbone, P.; Credi, A.; Di Stefano, M.;
Masiero, S.; Negri, F.; Orlandi, G.; Spada, G. P. J. Phys. Chem.
A
2006, 110, 12385ꢀ12394; (c) Uchida, E.; Sakaki, K.;
Nakamura, Y.; Azumi, R.; Hirai, Y.; Akiyama, H.; Yoshida, M.;
Norikane, Y. Chem. Eur. J. 2013, 19, 17391ꢀ17397; (d)
Schweighauser, L.; Häussinger, D.; Neuburger, M.; Wegner, H.
A. Org. Biomol. Chem. 2014, 12, 3371ꢀ3379.
7. Evans, M. G.; Polanyi, M., J. Chem. Soc., Faraday Trans.
1936, 32, 1333ꢀ1360.
8. Petersen, A. U.; Broman, S. L.; Olsen, S. T.; Hansen, A.
S.; Du, L.; Kadziola, A.; Hansen, T.; Kjaergaard, H. G.;
Mikkelsen, K. V.; Nielsen, M. B. Chem. Eur. J. 2015, 21, 3968ꢀ
3977.
9. For relationships between ring size, yield, and melting
points of cycloalkane and other macrocycles, see for example:
Dale, J. J. Chem. Soc. 1963, 93ꢀ111.
10. Lubrin, N. C. M.; Vlasceanu, A.; Frandsen, B. N.; Skov,
A. B.; Kilde, M. D.; Mikkelsen, K. V.; Nielsen, M. B. Eur. J. Org.
Chem. 2017, 2932ꢀ2939.
11. (a) Zheng, J.; Mielke, S. L.; Clarkson, K. L; Truhlar, D. G.
Comput. Phys. Commun. 2012, 183, 1803ꢀ1812; (b) Frisch, M.
J. et al., Gaussian, Inc., Wallingford CT, 2013.
12. (a) Poulsen, T. D.; Ogilby, P. R.; Mikkelsen, K. V. J.
Chem. Phys. 2001, 115, 7843ꢀ7851; (b) Kongsted, J.; Osted,
A.; Mikkelsen, K. V.; Åstrand, P.ꢀO.; Christiansen, O. J. Chem.
Phys. 2004, 121, 8435ꢀ8445.
ACS Paragon Plus Environment

Page 13 of 13
The Journal of Organic Chemistry
1
2
3
4
n = 0:
highest energy
storage
O
n
O
5
6
7
8
O
O
O
n
O
O
O
9
NC
CN
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
CN
O
O
n
NC
NC
O
O
CN
CN
Energy
discharging
NC
n = 0: ultraslow
n = 1: very slow
n = 2: slow
CN
CNNC
CN
fast
13
ACS Paragon Plus Environment