Intercomponent interactions and mobility in hydrogen-bonded rotaxanes
Jagesar, D.C.Publication date 2010
Link to publication
Citation for published version (APA):
Jagesar, D. C. (2010). Intercomponent interactions and mobility in hydrogen-bonded rotaxanes.
General rights
It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).
Disclaimer/Complaints regulations
If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.
C h a p t e r 6
Photoinduced Dynamics in
Structurally Modified
Naphthalimide Rotaxanes
AbstractThe photoinduced shuttling in a hydrogen-bonded rotaxane 11 was investigated. Compared to the imide rotaxane 1 described previously, which has a C12 saturated alkyl spacer between the succinamide (succ) and naphthalimide (ni) stations, 11 contains a glycine motif (Gly), incorporated in the thread next to the ni binding station combined with a C9 saturated alkyl spacer. The combined niGly station has a higher affinity for hydrogen bonding with the macrocycle than the ni station. Evidence for this is provided by the fact that in contrast to the other imide rotaxanes, both co-conformers of 11 are present in detectable amounts in solution. This was demonstrated by NMR and fluorescence experiments.
The photoinduced shuttling rate was found to be only slightly higher for 11 than for rotaxane 1. Shuttling in rotaxane 12, which only has a C9 spacer between succ and ni, however, was substantially faster. The temperature dependence of the shuttling rate implies that the moving ring in the niGly shuttle has to overcome a higher energy barrier before binding to the naphthalimide station. The greater energy barrier compared to 12 is caused by a more negative entropy of activation (
∆
S‡) associated with the shuttling process in 11. Based on the thermodynamic and kinetic characteristics of the shuttling process in the structurally similar rotaxanes, a mechanism for the shuttling process is proposed. In this model, the macrocycle translocation between the two stations occurs via an intermediate co-conformer in which the macrocycle is hydrogen-bonded with both stations.6.1
Introduction
The tuning of the performance of molecules and molecular assemblies with molecular motor functionality is one of the important focal points of the research on molecular motors.[1-6] Especially the macrocycle shuttling in rotaxanes is a subject of intensive research. The properties that are mainly targeted in these investigations are the efficiency,[7] selectivity[8-11] and rate[12-14] of the macrocycle translational motion.
One way to influence these properties is through external effects such as medium viscosity and polarity. For example, in Chapter 2 we demonstrated for the naphthalimide rotaxane 1 that the medium viscosity affects the shuttling rate and efficiency. Another strategy to manipulate the shuttling is by modifying the molecular structure, resulting in e.g. more favorable intercomponent interactions. Chapters 3 – 5 illustrated the possibility to tune the intercomponent hydrogen bonding in switchable rotaxanes by structural modification of the imide station. The studies with infrared spectroscopy revealed subtle differences in hydrogen bonding strength and geometry between the macrocycle and the mono- and dianions of the different imide stations.
In this chapter, two new structurally modified naphthalimide rotaxanes are presented. Rotaxane 11 (Figure 6-1) is similar to the previously described naphthalimide rotaxane 1, but three methylenes of the C12 alkyl chain are replaced by a glycine motif (Gly) incorporated in the thread next to the naphthalimide (ni) station. The Gly unit provides an additional amide group to which the macrocycle can bind. Because the glycine unit has approximately the same length as the (CH2)3 fragment, the distance between the succinamide (succ) and the naphthalimide station is practically the same as in 1, but the Gly motif could function as a stepping stone for shuttling. In this respect, the second binding station is effectively niGly and the distance to the succ station can be regarded as smaller than in rotaxane 1. Rotaxane 12 is similar to rotaxane 1, but the distance between the succ and ni stations is smaller. Compared to 11, rotaxane 12 has the same separation between the two binding stations, namely a C9 saturated alkyl spacer.
Rotaxane 11 was assembled through formation of the macrocycle around the thread 10
using a five-component clipping reaction developed in the research group of D.A. Leigh (see section 1.2.2). After the reaction workup, including chromatographic purification on silica, rotaxane 11 was obtained with a satisfactory yield of 30%.[15]
The arrangements of the C=O groups of the niGly and succ motifs are similar. In both cases, the two C=O groups are three bonds apart from each other and can adopt an anti-parallel conformation. This spatial arrangement of the C=O groups is favorable for the formation of two sets of bifurcated hydrogen bonds with the macrocycle.[16,17] Hence, the
niGly motif can be employed as template for the assembly of hydrogen bonded rotaxanes with the Leigh-type tetra amide macrocycles, although it may be less effective due to steric hindrance of the imide C=O group which is not involved in the template.
Figure 6-1 Structures of the niGly containing thread 10 and rotaxane 11, and the structurally related naphthalimide rotaxanes 1 (C12 alkyl spacer) and 12 (C9 alkyl spacer).
The research described in this chapter is mainly focused on rotaxane 11. The effect of the Gly unit in 11 on the co-conformer distribution and related photophysical behavior will be examined. Also, the shuttling dynamics in 11 will be discussed. The rotaxanes 1, 11 and
changes in the macrocycle shuttling dynamics. By linking the observed changes to the structural variations responsible for these differences in shuttling behavior, hopefully more insight into the mechanism of the shuttling process can be gained. For this purpose, comparative studies of the shuttling dynamics in 1, 11 and 12 will be performed.
6.2
Co-conformer Distribution
The macrocycle positioning in rotaxane 11 in CDCl3 was determined from the
1H NMR spectra in CDCl3 at 298 K (Figure 6-3). The signal of the methylene protons (Hd and He) of
the succinamide station is significantly broadened and show an upfield shift of ca. 1.1 ppm relative to the same signal in the thread 10, indicating that the macrocycle resides over the
succ station. The shielding is caused by the ring current effect of the macrocycle xylylene moieties. Remarkably, the upfield shift for 11 is smaller than for the naphthalimide rotaxane
1, where under the same conditions a shift of ∆δ = -1.45 ppm is observed.[18] For rotaxane 1
it was confirmed that the macrocycle is located exclusively at the succ station (see Chapter 3). The smaller upfield shift for rotaxane 11 can only mean that in this case not all molecules exist as the succ-11 co-conformer (Figure 6-2).
Figure 6-2 The co-conformers of rotaxane 11 resulting from macrocycle shuttling between the
succ and niGly station. The letters correspond to the assigned proton signals in the 1H NMR spectra of the thread and rotaxane in Figure 6-3.
Figure 6-3 Partial 1H NMR spectra of thread 10 and rotaxane 11 in CDCl3 at 298 K. The
assignment of the proton signals corresponds to the lettering shown in Figure 6-2.
The 1H NMR spectrum provides further evidence that a significant fraction of 11 is present as co-conformer niGly-11. This can be concluded from the upfield shift (0.3 ppm) of the methylene proton signal (Hl) of the Gly unit compared to the corresponding signal in
the thread. Also, the signal of the adjacent methylene protons (Hj) is shifted in the same
direction. The shielding of the protons of the niGly unit is obviously caused by the macrocycle in co-conformer niGly-11. The equilibrium between the two co-conformers was recently examined in our research group in detail by D. Günbaş with temperature dependent NMR experiments in CDCl3.
[15]
In the temperature range of 268 – 328 K a gradual increase from 10% to 27% of the niGly co-conformer population was observed. At room temperature (298 K) in CDCl3, 20% of 11 exists as the niGly co-conformer.
6.3
Photophysical Behavior
The absorption and emission spectra of thread 10 and rotaxane 11 in MeCN and CH2Cl2 are depicted in Figure 6-4 and the maxima are collected in Table 6-1; the data for rotaxane 1
are also given for comparison.[19] The absorption maxima of both molecules are slightly red-shifted (ca. 3 nm) in CH2Cl2 compared to those in MeCN. In both solvents, the spectra of the thread and rotaxane are almost identical. Also the fluorescence spectra of the thread and rotaxane are almost identical in both solvents. In addition, no differences are observed between the fluorescence spectra in MeCN and CH2Cl2. The absorption and emission spectra do not exhibit features attributable to the niGly-11 co-conformer. The absorption maximum of the rotaxane is slightly red-shifted (ca. 1 nm) compared to that of the thread, but the shift is too marginal to be considered significant.
(A) (B)
(C) (D)
Figure 6-4 Scaled absorption (A and C) and fluorescence (B and D) spectra of thread 10(····) and rotaxane 11 (—) in CH2Cl2 and MeCN. The fluorescence spectra were recorded with
excitation at 324 nm. The small irregularity in the absorption spectra at 350 nm is an artifact due to lamp switching in the spectrophotometer.
Table 6-1 Absorption and fluorescence maxima of thread 10, rotaxane 11 and rotaxane 1 in MeCN and CH2Cl2.
Compound λmax absorption (nm) λmax fluorescence (nm) MeCN CH2Cl2 MeCN CH2Cl2 Thread 10 339, 353 342, 356 379, 394 379, 396 Rotaxane 11 340, 354 343, 356 380, 395 380, 396 Rotaxane 1[a] 338, 352 - 376, 394 375, 395
[a] The fluorescence maxima for 1were taken from reference [19].
The fluorescence quantum yields (Φf) of 10 and 11 in MeCN and CH2Cl2 were measured by comparing the integrated emission across the band of the sample to that of a reference with known fluorescence quantum yield (quinine bisulfate in 0.1 M H2SO4).
[20] The obtained quantum yields for 10 and 11 are collected in Table 6-2; for comparison the values for the naphthalimide rotaxane 1 are also included. The Φf is significantly smaller in CH2Cl2 than in MeCN, both for the rotaxane and the thread. Also, in both solvents the quantum yields of the niGly rotaxane and thread are similar.
Table 6-2 Fluorescence quantum yields of thread 10 and rotaxane 11 in MeCN and CH2Cl2,
determined versus quinine bisulfate (λexc = 340 nm).
Compound Φf
MeCN CH2Cl2
Thread 10 0.30 0.23
Rotaxane 11 0.29 0.22
Rotaxane 1[a] 0.15 0.12
[a] The Φf values for 1 were taken from reference [19].
The Φf of 10 and 11 are much larger than for the naphthalimide rotaxane 1. Also, examination of the emission maxima (Table 6-1) reveals a slight red-shift of the highest-energy maximum of the niGly compounds compared to that of the naphthalimide rotaxane. These results can be explained by the presence of an internal hydrogen bond in the niGly unit (Scheme 6-1). The seven-membered ring geometry of this hydrogen-bonded conformer is comparable to those generally observed for diamides separated by three bonds, for example in the succinamide model compound 4 (Scheme 3-2).[21] The comparison of the fluorescence lifetimes of the niGly compounds and naphthalimide rotaxane 1 (see below) also indicates the presence of this hydrogen bond. The higher Φf for the niGly compounds show that the internal hydrogen bond in the niGly unit has a substantial effect on the non-radiative decay of the naphthalimide chromophore. Although in the rotaxane in a minor fraction this internal hydrogen bond is broken, it is replaced by hydrogen bonds between
the macrocycle and the chromophore, with a negligible net effect on the photophysical properties. The lowering of the S0 – S1 energy gap in the niGly compounds can also be attributed to the intramolecular hydrogen bonding in the niGly unit.
Scheme 6-1 Intramolecular hydrogen bonding in the niGly station.
The fluorescence lifetimes (τf) of the thread 10 and rotaxane 11 in MeCN and CH2Cl2 were measured by means of time correlated single photon counting (TCSPC). The solutions were excited at 324 nm and the fluorescence was detected at three different wavelengths: 365, 380 and 430 nm. The lifetimes obtained from fitting of the TCSPC data (with deconvolution) are summarized in Table 6-3.
Table 6-3 Fluorescence decay times of the niGly thread 10 and rotaxane 11 in MeCN and CH2Cl2 at three different detection wavelengths. The relative amplitudes of the components are
given in parentheses.
Solvent λdet Thread 10 (ns) Rotaxane 11 (ns)
(nm) τ1 τ2 τ1 τ2 τ3 MeCN 365 2.1 (0.91) 0.43 (0.09) 2.2 (0.65) 1.1 (0.19) 0.019 (-0.16) 380 2.1 (0.97) 0.77 (0.03) 2.2 (0.67) 1.3 (0.18) 0.026 (-0.15) 430 2.1 (0.91) 0.037 (-0.09) 2.2 (0.66) 1.3 (0.13) 0.036 (-0.21) CH2Cl2 365 1.5 (0.92) 0.43 (0.08) 1.5 (0.84) 0.58 (0.16) - 380 1.4 (0.86) 0.040 (-0.14) 1.5 (0.73) 0.70 (0.11) 0.038 (-0.16) 430 1.4 (0.82) 0.049 (-0.18) 1.5 (0.72) 0.73 (0.11) 0.064 (-0.17)
For the thread, the fluorescence follows a bi-exponential decay in both MeCN and CH2Cl2. The lifetime of the naphthalimide fluorescence in MeCN (τf = 2.1 ns) is longer than in CH2Cl2 (τf = 1.5 ns), which is consistent with the lower fluorescence quantum yield in the latter. The lifetimes for 10 and 11 in MeCN are longer than for the naphthalimide rotaxane
1 (in MeCN (τf = 1.5 ns).[19] This is also in agreement with the higher fluorescence quantum yields of the niGly compounds. The longer fluorescence lifetimes are also consistent with the previously drawn conclusion that the internal hydrogen bond in the niGly unit has a substantial effect on the non-radiative decay of the naphthalimide chromophore. In both
solvents, an additional decay component was found at short detection wavelengths. This component was not found at longer wavelengths. Instead, a short component (40 – 50 ps) with negative amplitude was observed. This component might be related to changes in hydrogen bonding between the Gly unit and naphthalimide chromophore in the excited state (Scheme 6-1). Calculations on 1,8-naphthalimides predict a larger electron density in the imide ring for the LUMO than for the HOMO.[22,23] Therefore, the hydrogen bond acceptor strength of the excited state is probably larger than that of the ground state and consequently, hydrogen bonding should be stronger in the excited state.[24]
The lifetime of the naphthalimide chromophore in rotaxane 11 is the same as that found in the thread 10. Interestingly, an additional short component of 1.1 – 1.3 ns in MeCN and 0.58 – 0.73 ns in CH2Cl2 is present. The relative amplitude is ca. 15% in both solvents. This component could be the fluorescence lifetime of the niGly-11 co-conformer. The shorter lifetime is then due to an increased rate of non-radiative decay induced by the presence of the macrocycle.
6.4
Photoinduced Shuttling
6.4.1niGly
Rotaxane11
In degassed solutions of 10or 11in acetonitrile, the characteristic T-T absorption of the
ni chromophore appears right after the end of the laser pulse (Figure 6-5) with maxima at 371, 447 and 471 nm; in the rotaxane the lowest-energy maximum is slightly red-shifted to 472 nm. These maxima are similar to those of the naphthalimide rotaxane 1.
Figure 6-5 Triplet-triplet absorption spectra of thread 10 (····) and rotaxane 11 (—) in MeCN, recorded 30 ns after the laser pulse (λexc = 355 nm).
The photoinduced shutting in rotaxane 11 was studied in MeCN, using the same photoactivation strategy as for the naphthalimide rotaxane 1 (Chapter 2, Scheme 2-1).[25] The phenomena observed upon laser-excitation of an MeCN solution containing 10 or 11
and the electron donor DABCO are similar to those observed for the naphthalimide shuttle
1. The triplet state is quenched by electron transfer from DABCO, and the characteristic absorption of the naphthalimide radical anion (ni●–
) grows in during the first 50 ns after laser-excitation. For thread 10 the absorption maximum is at 414 nm and does not change in time. Rotaxane 11 has a slightly blue-shifted maximum at 413 nm compared to the thread, and undergoes a blue-shift of ca. 3 nm during the first microseconds after laser-excitation. Just like in the naphthalimide shuttle 1, this blue-shift is direct evidence for macrocycle shuttling: it signals the formation of the energetically more favorable niGly-11●–
co-conformer. The blue-shift reflects the stabilization of the ground state of the naphthalimide radical anion by hydrogen bonds with the macrocycle. The rate of the blue-shift of the absorption maximum is directly linked to the rate at which co-conformer
succ-11●– is converted to niGly-11●–. This is a first-order process and can be described by a mono-exponential function. The shuttling rate was thus determined by fitting a mono-mono-exponential function to the plot of λmax versus time (Figure 6-6); the time constant of this fit is the shuttling rate constant kshuttle.
(A) (B)
Figure 6-6 (A): Transient absorption spectra of rotaxane 11 in MeCN in the presence of 10 mM DABCO recorded 130 ns after laser-excitation, and with subsequent 100 ns increments (black to grey). (B): position of the absorption maximum (λmax) of the radical anion as a function
of time. The line is the fit to a mono-exponential function.
It was concluded from the 1H NMR studies that in the neutral state a significant fraction of 11 already exists as the niGly-11 co-conformer (20% at room temperature in CDCl3).
[15]
This co-conformer is, however, not expected to interfere with the abovementioned method for the determination of the shuttling rate, because the conversion of niGly-11 to niGly-11●–
and its associated blue-shift will occur much faster than the conversion of succ-11●– to
niGly-11●– (microseconds). Therefore, the observed blue-shift of the radical anion absorption maximum can solely be attributed to shuttling. It should be noted that the absorption maximum of 11●–observed directly after its formation will be blue-shifted compared to that of 10●– because of the presence of niGly-11●– which is rapidly formed from niGly-11. This is indeed confirmed by the experimental result. The absorption maximum of 11●–, directly after its formation, is slightly blue-shifted to 413 nm compared to thread 10●–; the latter has a maximum at 414 nm.
The time constant of the shuttling process in 11●– obtained from the transient absorption experiments is kshuttle = (1.56 ± 0.15) × 106 s-1. This value is larger than that obtained for the
naphthalimide shuttle 1 under the same conditions (for 1 kshuttle = 1.30 × 10
6 s-1, see Chapter 2),[25] but the difference is small.
The driving force for the shuttling process is the larger hydrogen bonding affinity of the reduced imide station compared to that of the succinamide station. It was demonstrated previously that the hydrogen bond affinity of the imide station does influence the shuttling rate to some extent. This was done by comparison with shuttles containing a naphthalene diimide (ndi) or pyromellitimide (pmi) instead of the naphthalimide station of 1: both exhibit smaller shuttling rate constants.[26] Along the same lines we would expect that the additional amide group in 11, which can in fact act as a stepping stone for the shuttling by enhancing the hydrogen bond affinity of the combined niGly station, will also affect the shuttling rate. Despite the fact that the faster shuttling in 11 is consistent with this idea, it is however not clear whether the faster dynamics is caused by the greater hydrogen bond affinity or the shorter distance between the succ and niGly station.
6.4.2
C
9 Rotaxane12
In order to test the importance of the distance travelled by the macrocycle, and thereby to get more insight into the precise role of the Gly unit in the shuttling process, comparative experiments were performed with the naphthalimide rotaxane 12 (Figure 6-7B), which has the same binding stations as 1, but separated by a C9 instead of a C12 alkyl spacer. Based on the assumption that the Gly and ni units of 11 act as one single niGly station, the distance from the succ station is effectively the same in 11 and 12, because both contain a C9 alkyl spacer. Therefore, the shuttling rate in 12 is expected to be similar to that in 11 and a bit larger than in 1. The obtained shuttling rate constant in 12 is indeed larger than in 1, but to our surprise it is much larger than in 11: kshuttle = (3.3 ± 0.4) × 10
(A) (B)
Figure 6-7 (A): Shift of the absorption maximum of 12●– in MeCN (containing DABCO) as a function of time. The line is the fit to a mono-exponential function. (B): Shuttling in the radical anion of the C9 rotaxane 12. The blue-shift in (A) is due to the shuttling process (conversion of
succ-12●– to ni-12●–).
6.5
Temperature Dependence of the Shuttling Rate
The results of the comparative experiments with shuttles 1 and 11 suggested that the Gly unit of 11 might act as a stepping-stone for the macrocycle during the translocation from the succ to the ni station. In that case, the effective distance between the stations can be considered smaller than in shuttle 1. On the other hand, the significantly higher shuttling rate in 12, in which the length of the alkyl spacer between the two stations is the same as in11, suggests that the distance between the stations is not the (only) determining factor for the shuttling rate.
In order to get more insight into this issue, the temperature dependence of the shuttling process in 11 in MeCN was investigated. The shuttling rate constants (kshuttle) of 11 were determined at different temperatures in the range of 248 – 298 K. The Gibbs energy of activation ∆G‡, the activation enthalpy ∆H‡ and activation entropy ∆S‡ for the shuttling process were determined using the Eyring transition state theory (Eq. 6-1).[27] The Eyring activation parameters were obtained from the linear plot of ln (kshuttle/T) versus 1/T (Figure 6-8). The slope of the linear relationship equals –(∆H‡/R) and the intercept is the quantity ln(κ kB/h) + ∆S
‡/R. The relationship between the activation parameters ∆
G‡, ∆H‡, and ∆S‡ is expressed by Eq. 6-2. − = T R H R S h T k kshuttle B ‡ ‡ exp exp
∆
∆
κ
Eq. 6-1‡ ‡ ‡ S T H G
∆
∆
∆
= − Eq. 6-2 with:R : gas constant (8.31 J K-1 mol-1)
T : temperature (K)
κ : transmission coefficient (assumed equal to 1)
kB : Boltzmann’s constant (1.38 × 10
-23 J K-1)
h : Planck’s constant (6.626 × 10-34 J s)
The kshuttle at different temperature and the calculated activation parameters are collected
in Table 6-4 and Table 6-5, respectively. In Table 6-5 results are included from reference [29] and from reference [28], in which the shuttling process in 12 was studied by time-resolved infrared spectroscopy. In this study, the rate for 12 at 298 K was found to be kshuttle
= (3.6 ± 0.2) × 106 s-1, in good agreement with our result.
Table 6-4 Shuttling rate constants (kshuttle) for shuttle 11 in MeCN at different temperatures.
T (K) kshuttle (10 6 s-1)[a, b] 248 0.16 258 0.26 268 0.44 278 0.62 288 0.92 298 1.3
[a] Average of at least two measurements with the same sample. [b] The fitting error is approximately 10%.
Table 6-5 Eyring activation parameters for the shuttling process in rotaxanes 11, 1 and 12 in MeCN. Parameter 1[a] 11 12[b]
∆
G‡ (kcal mol-1) [c] 9.3 ± 1.0 9.1 ± 0.2 8.5 ± 0.8∆
H‡ (kcal mol-1 ) 7.5 ± 0.9 5.7 ± 0.1 6.2 ± 0.2∆
S‡ (cal mol-1 K-1) -6.3 ± 0.5 -11.4 ± 0.2 -7.6 ± 0.7[a] Data were taken from reference [29]. [b] Data were taken from reference [28]. [c] At 298 K.
6.6
Discussion
The Eyring analysis reveals that the relatively slow shutting in 11 is caused by a significantly more positive contribution of the T
∆
S‡ term (Eq. 6-2) to the∆
G‡ for 11 than for 1, i.e. due to a more negative∆
S‡ for11. The more negative
∆
S‡ suggests that in the transition state, the system is characterized by a larger extent of ordering in 11 than in 1. The∆
H‡ for11 is much smaller than for 1, but due to the more negative
∆
S‡, the∆
G‡ of the shuttling process in 11 is remarkably similar to that for shuttle 1, which explains the small difference in shuttling rates.The
∆
G‡ for the shuttling process in the C9 rotaxane 12 is lower than that in the niGlyrotaxane 11. Interestingly, the most relevant factor is the T
∆
S‡ contribution to∆
G‡: for 12,the
∆
S‡ is less negative than for 11. The possible mechanism responsible for the more negative∆
S‡ for11 is discussed below. Shuttling in the C9 rotaxane 12 is ca. 2.5 times faster
than in the C12 rotaxane 1. Therefore, at first sight it is somewhat surprising that the
∆
S‡ for shuttling in 12 is a bit more negative than for the process in 1. However, the enthalpy of activation is sufficiently low to compensate for the negative T
∆
S‡ contribution. The net result is a lower free energy of activation, hence the faster shuttling in 12 than in 1.The temperature-dependent shuttling experiments with the molecular shuttles described above delivered some puzzling results. In the initial working hypothesis by Wurpel for photoinduced shutting in the imide-based rotaxanes, it was stated that the activation barrier for the shuttling process is only determined by breaking of hydrogen bonds between the macrocycle and the succinamide station.[19,30] The main argument for this model was the fact that the shuttling rate increases if the hydrogen bonding is weaker, which occurs if the polarity of the solvent increases (in the series butyronitrile, propionitrile and acetonitrile). Eyring analysis revealed that the faster shuttling in MeCN compared to that in PrCN is caused by a less negative
∆
S‡ in MeCN.[29]This model was modified by Baggerman by explicitly treating the shuttling as a two-step process in which a metastable intermediate is formed.[26] This approach was used to explain the different shuttling rates in a series of rotaxanes containing different aromatic imide
stations (naphthalimide, naphthalene diimide and pyromellitimide). The first step in this refined model is the dissociation of the succ co-conformer: breaking of hydrogen bonds with the succ station and diffusion of the macrocycle from this station to form a metastable intermediate in which the macrocycle resides over the alkyl chain. In this metastable state, the macrocycle can either move back to the initial succ station or move forward towards the imide anion station. In the latter case the macrocycle is trapped on the imide anion station due to a lower free energy of the resulting co-conformer. The difference with the original hypothesis of Wurpel is that the intermediate is separated from both stable co-conformers by energy barriers. The probabilities of the trapping on either station depend on the heights of the barriers for both processes, whereas in Wurpel’s model trapping on the imide anion acceptor is supposed to be fast, and not rate limiting. An important factor that may contribute to a barrier for binding to the imide anion is the conformational change of the macrocycle from an initial chair to a boat-like conformation necessary for binding to this station (see for example Figure 6-7B). The energy barrier for this conformational change will be lower in solvents of high polarity because the boat conformation possesses a larger dipole moment.[26] This polarity effect may contribute to faster shuttling in acetonitrile than in butyronitrile. The spatial arrangement of the hydrogen bond accepting C=O groups (i.e. the structure of the imide anion station) also puts constraints on the conformation of the macrocycle, which may also contribute to the energy barrier for the shuttling process.
This mechanism does not explain the obtained activation parameters for 1, 11 and 12. According to this model the required rate-limiting conformational change of the macrocycle for binding to the imide station and therefore the free energy barrier of the shuttling process will be the same for 1 and 12. The experimental result, however, shows that these are different. Also, the reaction path as proposed by this barrier-controlled mechanism does not provide an explanation for the negative
∆
S‡, because the ordering in each stage of the process seems to be similar or even lower than in the firmly bound succ-co-conformer. Due to the existence of two barriers, the metastable intermediate could live long enough to allow movement of the macrocycle back and forth along the thread. Since this motion is faster than the overall shuttling process,[28,31] Wurpel’s model allows to attribute the distance dependence of the observed shuttling rate to the probability of a successful random walk, which decreases with increasing spacer length. Barriers between the intermediate state and the traps, however, lead to a loss of the distance dependence: if the intermediate state is sufficiently long-lived, the rate only depends on the barrier heights.In order to explain the observed shuttling behavior in the naphthalimide rotaxanes, we propose an alternative mechanism for this process. In this model, the shuttling is also treated as a two-stage process, but it occurs via an intermediate state in which the macrocycle is hydrogen-bonded to both stations. This so-called alkyl co-conformer intermediate (Figure 6-9) is then rapidly converted to the ni co-conformer. In essence, this model states that the macrocycle translocation is assisted by the imide radical anion, and can
therefore be denoted as a harpooning model. Such an alkyl co-conformer in which the macrocycle is simultaneously hydrogen-bonded to two distant amide stations, has been reported in the literature.[8] In that particular case, the C
12 alkyl spacer adopts a folded S-shape conformation in order to allow the amides on both sides of the spacer to reach the macrocycle for hydrogen bonding. The arrangement is relatively favorable because the stations are only poor hydrogen bond formers. We envisage that the alkyl co-conformer is formed from a conformation in which the alkyl chain folds such that hydrogen bonds between the imide anion and macrocycle can be formed. The conformation of the alkyl chain in this folded species can be compared to that in the folded-2●– conformation detected in THF for the radical anion of the naphthalimide thread (Scheme 3-3).
The structure of the transition state resembles that of the alkyl co-conformer. According to the transition state theory, the transition state corresponds to the maximum on the reaction coordinate. Therefore, it is best represented by a structure that has a geometry similar to that of the alkyl co-conformer but without hydrogen bonds between the macrocycle and the succ station (Figure 6-9). Thus, in the rate-limiting step there is a net reduction in the number of hydrogen bonds, in agreement with the explanation for the solvent effect proposed by Wurpel et al.[19]
Figure 6-9 Proposed model of the shuttling process in the imide-based molecular shuttles. Schematic representation of the free energy profile of the different stages of the shuttling process, and the corresponding co-conformers.
The negative entropy of activation for the shuttling process can be explained in the context of this model. Binding to both stations reduces the conformational degrees of freedom of the shuttle. Especially for the alkyl spacer a significant number of possible rotational isomers are lost. The distance between the two stations (length of the alkyl spacer) is therefore expected to have an effect on the entropy of activation. It is not difficult
to imagine that the alkyl co-conformer will be formed at a lower entropic cost if the distance between the two stations is smaller. Because of the shorter alkyl spacer, the shuttling process in 12 (C9 spacer) is expected to have a less negative
∆
S‡ compared to that in 1 (C12 spacer). The experimental result shows that this is not the case: the
∆
S‡ for 12 is slightly more negative than for 1. This might be due to a difference in conformational freedom of the macrocycle in the transition states of 1 and 12, which leads to enthalpy gain at the expense of entropy. The phenomenon of entropy-enthalpy compensation is not uncommon in processes where weak interactions play a dominant role.[32,33] The reason is that some freezing of the conformational freedom is required in order to meet the conditions associated with the enthalpic contribution and consequently
∆
S‡ will become more negative. In the present case one could intuitively imagine that in the transition state hydrogen-bonding with the imide station will become easier if the alkyl spacer becomes shorter (due to less steric hindrance by the alkyl spacer), giving rise to a lower∆
H‡. As a consequence, because of the firmer binding, the macrocycle and alkyl chain in the transition state will be more organized and the∆
S‡ will become more negative. The smaller∆
H‡ for12
compared to 1 indeed implies the existence of stronger macrocycle-imide hydrogen-bonding interactions in the transition state of 12. Consequently, the reduction of the degrees of freedom of the macrocycle might be stronger in 12 than in 1. It was argued above that the
∆
S‡ is negative, primary because of folding of the alkyl chain, but it is likely that the restricted conformational freedom of the macrocycle also gives rise to a negative contribution to the∆
S‡.For the mechanism of the shuttling process in the niGly shuttle 11, two different scenarios are imaginable. The first possibility is that, because of the much stronger hydrogen bond affinity of the radical anion, the Gly unit is not involved in the shuttling process. In this case, the macrocycle directly migrates from the succ to the ni●–
station. Thus, the distance between the succ and ni●–
stations in the shuttles 11 and 1 will be effectively the same and similar activation parameters might be expected. The second possibility is that the
Gly unit acts as a stepping stone for the shuttling process i.e. the Gly unit assists the shuttling by binding to the macrocycle before the latter binds to the imide radical anion. The different activation parameters of 11 compared to those of 1 are a clear indication that shuttling in 11 occurs via a mechanism in which the binding in the transition state is different than in 1.
The
∆
H‡ value for shuttling in11 is smaller than that in 1, but also slightly smaller than that in 12, indicating stronger hydrogen bonding in the transition state of 11. Nevertheless, the shuttling in 12 is much faster than in 11. The reason for this is the very negative
∆
S‡ for the shuttling in 11, which implies that the transition state of 11 is more organized than that of 1 or 12. In the alkyl-11 co-conformer the macrocycle can be hydrogen-bonded to the Gly or the ni●–unit, or to both. In Figure 6-10 several possible geometries for the alkyl co-conformer are sketched.
Figure 6-10 Three possible structures of the alkyl co-conformer of 11●–
. In all three cases, hydrogen bonds exist with one amide group of the succinamide. In (A) the other end of the ring hydrogen-bonds to the Gly C=O, in (B) it binds to the imide C=O and in (C) it binds to both. Note that many other structures of this kind are conceivable, because hydrogen bonding of the C=O groups of the macrocycle with the NH groups of the succ and Gly units are also possible.
The structures A, B and C have different extents of organization and different energies. The hydrogen bonding pattern in structure B is the same as might be expected in the alkyl co-conformer of shuttle 1. Because of the more negative
∆
S‡ for the shuttling in 11, we can therefore conclude that structure B is probably not the intermediate co-conformer in the shuttling process in 11. Also, structure A can be rejected because the ordering is even lower than in structure B. The most likely candidate for the structure of the intermediate alkyl-11●–is structure C. This species is more organized than structures A and B, because hydrogen bonding with both the ni●–
and Gly unit further reduces the conformational degrees of freedom of the alkyl spacer, the macrocycle and the Gly unit. This analysis shows that the
Gly unit does not act as a stepping stone, a separate binding site nine methylenes away from the succ station. Instead, it assists the shuttling process by lowering of the
∆
H‡ in a transition state in which the ring is already bound to the ni station. But this does not result in much faster shuttling because its action is also associated with a more negative∆
S‡.The proposed intermediate alkyl co-conformer, from which the transition state is formed, is expected to be short-lived, due to conversion to the energetically more favorable
ni co-conformer or, alternatively, the shuttle can return to its initial succ co-conformer. Due to its short lifetime, the direct detection of the intermediate alkyl co-conformer would require a time-resolved spectroscopic technique with (sub)nanosecond time resolution. An attractive method for this purpose would be time-resolved infrared spectroscopy. The alkyl conformer is expected to possess different spectroscopic signatures compared to the co-conformers in which the macrocycle is only hydrogen-bonded to either the succinamide or the imide radical anion. The detection of the alkyl co-conformer might be possible, because hydrogen bonds of the macrocycle with both the succ and the ni●–
red-shifts of the stretching frequencies of the involved C=O and N-H groups of both stations. Such experiments with a time resolution of 10 ns are described in reference [28] for rotaxanes 1 and 12 and two analogs with shorter (C5) and longer (C16) spacers. The shuttles were activated using the same approach as described in the present work. The authors describe the shuttling as a two-stage process: detachment of the macrocycle from the succ station followed by a biased random walk towards the imide anion station. The time constant of the departure of the macrocycle from the succinamide station was derived from the decay of the C=O stretching band of the hydrogen-bonded succinamide, and the rise of free C=O stretching band in this station. The time constant of the arrival of the macrocycle at the imide anion station was derived in a similar way by analysis of the free and hydrogen-bonded C=O stretching bands of this station. The departure and arrival dynamics were found to be the same (0.77 µs), which implies that the succ co-conformer is directly (in one step) converted into the ni●–
co-conformer. The IR spectra do not indicate the presence of intermediate states. This is not surprising because the concentration of this short-living alkyl co-conformer will be very low. It may be impossible to detect the intermediate state because the IR absorptions of this species have to be filtered from a background of absorptions of the succ and ni●–
co-conformers.
6.7
Conclusion
The macrocycle shuttling dynamics in the hydrogen-bonded rotaxane 11, containing a succinamide (succ) and a naphthalimide-glycine (niGly) binding station were investigated. The additional Gly close to the naphthalimide turned out to have a significant impact on several properties of the rotaxane.
The hydrogen bond affinity of the niGly station is strengthened significantly compared to that of the naphthalimide alone. In the latter case (e.g. in 1 and 12) the co-conformer equilibrium lies completely on the side of the succ co-conformer. In the neutral state of 11, a significant fraction exists as the niGly co-conformer. This is evident from the 1H NMR spectrum: the presence of the macrocycle at the niGly station is revealed by an upfield shift (0.3 ppm) of the signal of the glycine methylene protons, compared to the same signals in the thread. The favorable hydrogen bond affinity of the niGly station makes it an attractive template for the assembly of hydrogen-bonded rotaxanes.
Weak signatures of the niGly-11 co-conformer were found in the photophysical behavior. The fluorescence decay of the naphthalimide chromophore in the rotaxane contains an additional shorter lifetime which may be assigned to the niGly-11 co-conformer. The shorter lifetime of the niGly-11 co-conformer is attributed to enhanced non-radiative decay due to the presence of the macrocycle. Also, for both the thread and the rotaxane, a growing in of a short component (40 – 50 ps) was detected, which indicates the formation of internal hydrogen bonds in the niGly station in the excited state.
The shuttling rate in niGly rotaxane 11 is only 1.2 times larger than in the rotaxane 1
which lacks the Gly unit, but has the same distance between the succ and ni stations. The shuttling dynamics in 11 was also compared with that in C9-rotaxane 12, which has the same alkyl chain length between the stations as 11 but a ni instead of a niGly as the second station. Surprisingly, the shuttling in 12 is more than 2 times faster than in 11. In order to explain these results, a model for the shuttling mechanism is proposed in which shuttling takes place via an intermediate alkyl co-conformer. In this intermediate, the macrocycle resides over the alkyl chain between the two stations and is hydrogen-bonded to both stations. The structure of the transition state is similar to that of the alkyl co-conformer. It has nearly the same geometry, but the macrocycle is hydrogen-bonded to only the imide radical anion. Both the
∆
H‡ and the∆
S‡ of the shuttling process are affected by the spacer length. Hydrogen bonding with both stations has an unfavorable effect on the∆
G‡, because the∆
S‡ is negative due to lowering of conformational degrees of freedom of the alkyl chain and the macrocycle. Therefore, a long alkyl spacer will contribute more to the T∆
S‡ term of∆
G‡(Eq. 6-2) than a short spacer. At the same time, hydrogen bonding in the alkyl co-conformer will be stronger for short spacers (due to less steric hindrance), leading to a lower
∆
H‡. Strong hydrogen bonding (low∆
H‡) also means that the alkyl chain and macrocycle are more organized in the transition state, which will further lower the∆
S‡. A systematic study of the effect of the spacer length could provide more insight into this interplay between the∆
H‡ and∆
S‡.The faster shuttling in 12 compared to 1 is associated with a lower
∆
H‡ because of the shorter distance between the succ and ni stations and thus stronger hydrogen bonding in the transition state. The∆
H‡ for11 is lower than for 1 and 12 but
∆
S‡ is much more negative. The reason for this is that in the transition state of 11 the macrocycle is hydrogen-bonded to both the Gly and the ni●–units. The Gly unit does not serve as stepping-stone, effectively shortening the distance for the shuttling. It assists the shuttling because it lowers the
∆
H‡through stronger binding between the macrocycle and the ni●–
Gly station. However, the effect of the enthalpy gain on the rate is largely cancelled because of a more negative
∆
S‡associated with its action.
6.8
Experimental Details
Materials
Compounds 10 and 11 were synthesized by Dr. Leszek Zalewski. Rotaxane 12 was synthesized by Dr. Bert Bakker. Spectroscopic grade solvents were used after distillation from calcium hydride.
UV-Vis Absorption and Fluorescence
Ground state electronic absorption spectra were recorded in quartz cuvettes (optical path length 1 cm) on a double beam Varian Cary 3E Spectrophotometer or on a single beam HP 8453 diode array spectrophotometer. Steady state fluorescence spectra were recorded on a Spex Fluorolog 3 system. This spectrometer was equipped with double monochromators in the excitation and emission channels. The excitation light source was a 450 W Xe lamp and the detector was a Peltier cooled R636-10 Hamamatsu photomultiplier tube. The emission was detected in a right-angle geometry.
The fluorescence quantum yields were measured by comparing the integrated emission (I) across the band of the sample to that of a reference (quinine bisulfate in 0.1 M H2SO4) with known fluorescence quantum yield (Φref). The absorbance (A) and integrated emission of the sample and reference were measured for 5 different concentrations. The concentrations were chosen such that the absorbance at the excitation wavelength (λexc = 340 nm) was below 0.1 in all cases. The slopes of plots of I versus the absorption factor (1-10-A) for the sample (asample) and the reference (aref) were used to calculate the quantum yield
(Φsample) with Eq. 6-3:
ref ref sample sample 0 n n a a Φ Φ = 22 Eq. 6-3
In this equation, n and n0 represent the refractive indices of the solvent of the sample and
reference, respectively. The fluorescence quantum yield of quinine bisulfate in 0.1 M H2SO4 was not available, therefore the value of 0.546 in 0.5 M H2SO4 was used.
[20]
The sample and reference solutions were excited at 340 nm.
Time-Correlated Single-Photon Counting
Fluorescence decays were recorded with a time-correlated single-photon counting (TCSPC) setup. The excitation source (324 nm) was a frequency doubled cavity dumped dye laser (Coherent model 700) pumped by a mode-locked Ar+ laser (Coherent 486 AS Mode Locker, Coherent Innova 200 laser). The fluorescence was detected with a microchannel plate detector (Hamamatsu R3809). In order to exclude polarization effects, a polarizer at magic angle (54.7º) with respect to the polarization of the laser beam was used in front of the detector. The instrument response function was recorded with the Raman scattering of water at 363.5 nm. Fitting was performed with routines implemented in Igor Pro.
Transient Absorption and Temperature Control
The samples for nanosecond-to-microsecond transient absorption experiment were degassed by applying three freeze-pump-thaw cycles. The spectra were obtained with the transient absorption setup described in experimental section 2.5. Transient absorption
experiments at different temperature were performed in a liquid nitrogen cooled optical cryostat (DN1704, Oxford Instruments) with a feedback controller unit (ITC4, Oxford Instruments).
Acknowledgements
The synthesis and NMR characterization of 10 and 11 were performed by Dr. Leszek Zalewski. Rotaxane 12 was synthesized by Dr. Bert Bakker. Their contributions are gratefully acknowledged.
6.9
References
[1] Balzani, V.; Credi, A.; Raymo, F. M. and Stoddart, J. F. Artificial Molecular Machines.
Angew. Chem. Int. Edit. 2000, 39, 3349-3391.
[2] Dietrich-Buchecker, C.; Jimenez-Molero, M. C.; Sartor, V. and Sauvage, J. P. Rotaxanes and Catenanes as Prototypes of Molecular Machines and Motors. Pure Appl. Chem. 2003, 75, 1383-1393.
[3] Feringa, B. L. The Art of Building Small: From Molecular Switches to Molecular Motors. J.
Org. Chem. 2007, 72, 6635-6652.
[4] Kay, E. R.; Leigh, D. A. and Zerbetto, F. Synthetic Molecular Motors and Mechanical Machines. Angew. Chem. Int. Edit. 2007, 46, 72-191.
[5] Kay, E. R. and Leigh, D. A. Beyond Switches: Rotaxane- and Catenane-Based Synthetic Molecular Motors. Pure Appl. Chem. 2008, 80, 17-29.
[6] Balzani, V.; Credi, A. and Venturi, M. Light Powered Molecular Machines. Chem. Soc. Rev. 2009, 38, 1542-1550.
[7] Balzani, V.; Clemente-Leon, M.; Credi, A.; Ferrer, B.; Venturi, M.; Flood, A. H. and Stoddart, J. F. Autonomous Artificial Nanomotor Powered by Sunlight. Proc. Natl. Acad. Sci.
U. S. A. 2006, 103, 1178-1183.
[8] Bottari, G.; Dehez, F.; Leigh, D. A.; Nash, P. J.; Pérez, E. M.; Wong, J. K. Y. and Zerbetto, F. Entropy-Driven Translational Isomerism: A Tristable Molecular Shuttle. Angew. Chem.
Int. Edit. 2003, 42, 5886-5889.
[9] Deng, W. Q.; Flood, A. H.; Stoddart, J. F. and Goddard, W. A. An Electrochemical Color-Switchable RGB Dye: Tristable [2]Catenane. J. Am. Chem. Soc. 2005, 127, 15994-15995. [10] Qu, D. H.; Wang, Q. C.; Ma, X. and Tian, H. A [3]Rotaxane with Three Stable States That
Responds to Multiple-Inputs and Displays Dual Fluorescence Addresses. Chem. Eur. J. 2005, 11, 5929-5937.
[11] Trabolsi, A.; Khashab, N.; Fahrenbach, A. C.; Friedman, D. C.; Colvin, M. T.; Cotí, K. K.; Benítez, D.; Tkatchouk, E.; Olsen, J. C.; Belowich, M. E.; Carmielli, R.; Khatib, H. A.; Goddard, W. A.; Wasielewski, M. R. and Stoddart, J. F. Radically Enhanced Molecular Recognition. Nat. Chem. 2010, 2, 42-49.
[12] Ghosh, P.; Federwisch, G.; Kogej, M.; Schalley, C. A.; Haase, D.; Saak, W.; Lutzen, A. and Gschwind, R. M. Controlling the Rate of Shuttling Motions in [2]Rotaxanes by Electrostatic Interactions: A Cation as Solvent-Tunable Brake. Org. Biomol. Chem. 2005, 3, 2691-2700.
[13] Katz, E.; Baron, R.; Willner, I.; Richke, N. and Levine, R. D. Temperature-Dependent and Friction-Controlled Electrochemically Induced Shuttling Along Molecular Strings Associated with Electrodes. ChemPhysChem 2005, 6, 2179-2189.
[14] Hirose, K.; Shiba, Y.; Ishibashi, K.; Doi, Y. and Tobe, Y. A Shuttling Molecular Machine with Reversible Brake Function. Chem. Eur. J. 2008, 14, 3427-3433.
[15] Günbaş, D. D.; Zalewski, L. and Brouwer, A. M. Energy Landscape of a Hydrogen-Bonded Non-Degenerate Molecular Shuttle. Chem. Commun. 2010, 46, 2061-2063.
[16] Gatti, F. G.; Leigh, D. A.; Nepogodiev, S. A.; Slawin, A. M. Z.; Teat, S. J. and Wong, J. K. Y. Stiff, and Sticky in the Right Places: The Dramatic Influence of Preorganizing Guest Binding Sites on the Hydrogen Bond-Directed Assembly of Rotaxanes. J. Am. Chem. Soc. 2001, 123, 5983-5989.
[17] Altieri, A.; Bottari, G.; Dehez, F.; Leigh, D. A.; Wong, J. K. Y. and Zerbetto, F. Remarkable Positional Discrimination in Bistable Light- and Heat-Switchable Hydrogen-Bonded Molecular Shuttles. Angew. Chem. Int. Edit. 2003, 42, 2296-2300.
[18] Altieri, A.; Gatti, F. G.; Kay, E. R.; Leigh, D. A.; Martel, D.; Paolucci, F.; Slawin, A. M. Z. and Wong, J. K. Y. Electrochemically Switchable Hydrogen-Bonded Molecular Shuttles. J.
Am. Chem. Soc. 2003, 125, 8644-8654.
[19] Wurpel, G. W. H. PhD Thesis, University of Amsterdam, Amsterdam, 2001, 105-130, http://dare.uva.nl/en/record/94487.
[20] Eaton, D. F. Reference Materials for Fluorescence Measurement. Pure Appl. Chem. 1988, 60, 1107-1114.
[21] Gellman, S. H.; Dado, G. P.; Liang, G. B. and Adams, B. R. Conformation-Directing Effects of a Single Intramolecular Amide-Amide Hydrogen-Bond - Variable-Temperature NMR and IR Studies on a Homologous Diamide Series. J. Am. Chem. Soc. 1991, 113, 1164-1173.
[22] Jacquemin, D.; Perpète, E. A.; Scalmani, G.; Frisch, M. J.; Ciofini, I. and Adamo, C. Fluorescence of 1,8-Naphthalimide: A PCM-TD-DFT Investigation. Chem. Phys. Lett. 2007,
448, 3-6.
[23] Miao, L. F.; Yao, Y. L.; Yang, F.; Wang, Z. D.; Li, W. and Hu, J. M. A TDDFT and PCM-TDDFT Studies on Absorption Spectra of N-Substituted 1,8-Naphthalimides Dyes. J. Mol.
Struct. Theochem 2008, 865, 79-87.
[24] Baggerman, J.; Jagesar, D. C.; Vallée, R. A. L.; Hofkens, J.; De Schryver, F. C.; Schelhase, F.; Vögtle, F. and Brouwer, A. M. Fluorescent Perylene Diimide Rotaxanes: Spectroscopic Signatures of Wheel-Chromophore Interactions. Chem. Eur. J. 2007, 13, 1291-1299.
[25] Jagesar, D. C.; Fazio, S. M.; Taybi, J.; Eiser, E.; Gatti, F. G.; Leigh, D. A. and Brouwer, A. M. Photoinduced Shuttling Dynamics of Rotaxanes in Viscous Polymer Solutions. Adv.
Funct. Mater. 2009, 19, 3440-3449.
[26] Baggerman, J. PhD Thesis, University of Amsterdam, Amsterdam, 2006, 95-131, http://dare.uva.nl/en/record/191667.
[27] Isaacs, N. Physical Organic Chemistry, Longman Group UK Limited, Essex, 1995, 87-125. [28] Panman, M. R.; Bodis, P.; Shaw, D. J.; Bakker, B.; Newton, A. C.; Kay, E. R.; Brouwer, A.
M.; Buma, W. J.; Leigh, D. A. and Woutersen, S. Operation Mechanism of a Molecular Machine Revealed Using Time-Resolved Vibrational Spectroscopy. Science 2010, 328, 1255-1258.
[29] Fazio, S. M. PhD Thesis, University of Amsterdam, Amsterdam, 2005, 15-36, http://dare.uva.nl/en/record/162883.
[30] Brouwer, A. M.; Frochot, C.; Gatti, F. G.; Leigh, D. A.; Mottier, L.; Paolucci, F.; Roffia, S. and Wurpel, G. W. H. Photoinduction of Fast, Reversible Translational Motion in a Hydrogen-Bonded Molecular Shuttle. Science 2001, 291, 2124-2128.
[31] Socol, M.; Zoon, P. D. and Brouwer, A. M. Unpublished Results.
[32] Houk, K. N.; Leach, A. G.; Kim, S. P. and Zhang, X. Y. Binding Affinities of Host-Guest, Protein-Ligand, and Protein-Transition-State Complexes. Angew. Chem. Int. Edit. 2003, 42, 4872-4897.
[33] Schneider, H. J. Binding Mechanisms in Supramolecular Complexes. Angew. Chem. Int. Edit. 2009, 48, 3924-3977.
