So far, the source of asymmetry in all HTI-based molecular motors is a sulfoxide with point chirality instead of a carbon stereogenic center to control directionality.23–27, 42, 43 For HTI motor 1, we have now changed the asymmetry source to a carbon based stereocenter, which resides on the indanone- in place of the thioindigo fragment. This stereogenic center invokes an alcohol moiety in close proximity to the central double bond – the axis of directional rotation – which can form an intramolecular hydrogen bond to the stator fragment. The sulfur atom at the thioindigo fragment is not oxidized, which is expected to lead to a notable red-shift of the absorption of the motor as compared to the related structural setup of first generation HTI-based motors. A bromide is introduced at the thioindigo fragment for the dual purpose of facilitating synthesis (as halogenated benzothiophenones are more stable and less prone to form thioindigo as side product) and to introduce a convenient handle for late-stage functionalization and applications.
A theoretical analysis on the DFT level of theory was conducted first to elucidate the motor rotation mechanism of 1. A direct comparison was made to the behavior of ether-derivative 2 lacking hydrogen bonding capacity. For clarity only structures with (R)-configured stereocenter are considered in the following. The theoretical description of 1 reveals a very unusual yet very well-suited energy surface for proper unidirectional light powered motor rotation (Fig. 2). The global minimum structure A-1 possesses E configuration of the central double bond and allows intramolecular hydrogen bonding between the hydroxy group and the carbonyl function of the thioindigo fragment. A (P)-helicity is adopted in this structure, which brings the larger i-propyl substituent at the stereogenic center to the same side as the methoxy-group in ortho-position with regard to the double bond. With the helicity already hinting at the rotation direction, photoisomerization in a counterclockwise manner would lead to population of isomer B-1 with Z configured double bond and (M)-helicity. Now the hydroxy group at the stereocenter and the ortho-methoxy group are placed at the same side of the molecule. During this photoconversion intramolecular hydrogen bonding with the carbonyl is disrupted. In the calculation no significant hydrogen bonding interaction between the hydroxy group as donor and the sulfur atom as acceptor are found. A chalcogen bonding interaction between the sulfur and the carbonyl oxygen, as recently observed in heterocylic HTI photoswitches,44 is also not likely in this case (see Supporting Information for details). Isomer B-1 possesses a very high energy of 7.55 kcal/mol according to theory and stabilizes in a thermal step by inverting helicity. This helix inversion continues the directional rotation and serves as ratcheting step in the motor mechanism. A transition state with 9.6 kcal/mol elevated energy is predicted for this process. As a result, isomer C-1 with Z-configuration of the double bond and (P)-helicity is formed. The energy of isomer C-1 is predicted to be lowered by 4.43 kcal/mol as compared to B-1, which leads to complete conversion at ambient to high temperatures. Still, the stabilized isomer C-1 possesses a significantly higher energy than the global minimum by 3.7 kcal/mol. Taken together, these values belong to the highest amounts of photoenergy being stored in metastable HTI switches and motor systems so far. The significant energy content would lead to a complete thermal conversion from C-1 to isomer A-1 at elevated temperatures. The sequence of isomer interconversions from A to B to C represents the first 180° unidirectional rotation of motor 1. We expected a similar two-step interconversion between three isomers for the second 180° rotation instead of the population of an additionally and highly unusual epoxide-intermediate D-1. Photoirradiation of C-1 would thus continue rotation in a counterclockwise sense leading to population of metastable isomer E-1 with E-configured double bond and (M)-helicity. Intramolecular hydrogen bonding would be reestablished in this step with the carbonyl as acceptor. In E-1 again the smaller hydroxy group at the stereocenter and the ortho-methoxy group are located on the same side of the molecule. Isomer E-1 is calculated to be 2.3 kcal/mol higher in energy compared to A-1 and could thus be converted completely into the latter by another thermal helix inversion step. However, we found in the calculations, that in this process intramolecular hydrogen bonding is retained and the predicted transition state is only 1.8 kcal/mol higher in energy. With such low transition state energy an experimental observation of the thermal E-1 to A-1 conversion would not be possible without resorting to ultrafast transient spectroscopy methods. In contradiction to this initial theoretical assessment, we found a conveniently stable intermediate experimentally, which decayed completely into A-1 with an associated Gibbs energy of activation of 13.0 kcal/mol. Since also experimental spectroscopy evidences (see below) and theoretical spectra did not match for the expected isomer E-1, we returned to a more comprehensive theoretical assessment. Two options were taken into account guided by the experimental observation that UV/vis absorption is strongly hypsochromically shifted. To explain this behavior we assumed that conjugation via the central double bond is broken in the experimentally observed intermediate and either an epoxide or a furane-based structure is formed. As is described in more detail in the Supporting Information, the calculated epoxide structure D-1 matches very well both energetically and spectroscopically with the experimentally observed intermediate. D-1 is formed by a conjugated intramolecular epoxidation reaction while at the same time the hydrogen bond donor shifts to the former carbonyl – and now enol - site of the thioindigo fragment. A strong intramolecular hydrogen bond is formed between the enol and the epoxide, locking the structure in a nearly perpendicular arrangement of the indanone rotor versus the thioindigo stator. D-1 thus represents a “trapped” 90° rotation intermediate possessing an exceptional high energy of 12 kcal/mol. This large portion of energy is directly stored from the incident light irradiation and could thus be expended to do actual work in the thermally activated follow up steps. It establishes an unprecedented supercharging of a light-driven molecular motor. The next thermally activated step is calculated to be an epoxide opening and proton transfer step, which is tied to further rotation to reach intermediate E-1. The initial rotation direction is maintained in this step and a considerable Gibbs energy of activation of 14.9 kcal/mol is predicted. The final E-1 to A-1 thermal helix inversion step was found in the initial calculations as described above. With the theoretically predicted energy profile a quantitative directionality is expected for motor 1 even at higher temperatures. Low temperature experiments should allow to observe four different isomers A-1, B-1, C-1, and D-1 and thus evidence unidirectionality directly.
To complete the theoretical analysis NMR, UV/vis as well as ECD spectra were calculated at the CAM-B3LYP/6-311G(d,p) level of theory and the effect of different solvents (THF or Et2O) were included through the polarizable continuum model (PCM) (see Supporting Information for details). The hydroxy group proton signal in the 1H NMR spectrum was found to be especially sensitive to the particular isomeric state and thus was predicted to be a powerful probe for conformational analysis. UV/vis spectra report on significant electronic changes such as E or Z configuration of the central double bond and especially its braking during the constitutional alteration step. Electronic circular dichroism (ECD) spectra are highly sensitive to the particular helicity of the isomers, which would complete an unambiguous assignment of spectral species in the experiments.
A similar theoretical analysis was conducted for methoxy-substituted derivative 2 lacking the capacity for intramolecular hydrogen bonding. In this case the energy profile reveals that compound 2 also is a motor but adheres to the classical four-step cycle (including states A-2, B-2, C-2, and E-2) with alternating photochemical and thermal helix inversion steps. The sense of directionality is the same as in motor 1, which directly shows that hydrogen bonding in this classical rotation mechanism is not responsible for unidirectionality. Additionally, the C-2 isomer is now the global minimum and significantly less energy is stored in the metastable states B-2 (3.13 kcal/mol) and E-2 (4.84 kcal/mol). With this energy profile full unidirectionality would still be present, as the thermal helix inversion steps from B-2 to C-2 and E-2 to A-2 still lead to complete conversions and thus to effective ratcheting. Like for 1 the most stable structures A-2 and C-2 feature a syn-relation of the large i-propyl group and the aromatic methoxy group in ortho-position to the central double bond at the five membered indanone ring. It thus becomes clear that the size differences at the stereogenic center are primarily responsible for dictating the sense of directionality.
Synthesis of motor 1 is described in detail in the Supporting Information and starts from commercially available 3-(2,5-dimethoxyphenyl)propanoic acid (3). After intramolecular Friedel-Crafts acylation the corresponding 4,7-dimethoxy-indanone (4) is obtained to which the i-propyl group was introduced via alpha-deprotonation and substitution with i-propyl iodide. The resulting indanone 5 was obtained in 31% yield. To our surprise prolonged treatment of 5 with base led to formation of the alpha-hydroxylated indanone 6. A similar reaction has been reported just when writing this manuscript by Crespi, Feringa, and co-workers under Lewis-acid conditions,41 which we were not aware of at the time when synthesizing our motor. After optimization a satisfactory yield of 60% was achieved when bubbling air through the solution to increase the dioxygen reactant level. A future closer examination of this reaction will reveal details about the mechanism and probe its scope, however it is already evident that the hydroxy group stems from the air’s dioxygen. Condensation of indanone 6 and benzothiophenone 7 in the presence of BCl3 gave the final motor 1 in 23% yield. Single crystals suitable for X-ray analysis could be obtained for the two stable isomers, the global minimum E-isomeric A-1 as well as for metastable Z-isomeric C-1, directly evidencing the molecular structure and intramolecular hydrogen bonding in A-1 (Fig. 3a). Deuteration experiments allowed to identify the 1H NMR signals of the OH proton in the different isomers of motor 1 directly (Fig. 3b).
After successful synthesis the thermal behavior of motor 1 was scrutinized first. At ambient temperature only two isomers could be observed and fully characterized, A-1 and C-1, as predicted by theory. The metastable isomer C-1 was isolated after irradiation of A-1 with 450 nm light and flash column chromatography separation of the two isomers. Upon prolonged heating to 50°C in CDCl3 solution C-1 is completely converted into A-1. When conservatively assuming that remaining 5% of C-1 cannot be observed in the 1H NMR experiment, the resulting equilibrium constant K = 95/5 at 50°C can then be translated into the corresponding Gibbs energy difference between A-1 and C-1 ΔG = 1.9 kcal/mol. This value is the lower limit of the energy difference between the two isomers and thus the theoretically predicted value of 4.86 kcal/mol difference is supported by experiment. A corresponding Gibbs energy of activation of ΔG‡ = 25 kcal/mol was determined from kinetic analysis of the thermal C-1 to A-1 conversion.
Next irradiation experiments were conducted at low temperatures to evidence the direct photoproducts formed upon irradiation of A-1 and C-1. Starting from A-1, in situ irradiation at − 130°C in a 1:1 mixture of THF-d8 : CS2 revealed the formation of one new isomer as the direct photoproduct first and subsequently formation of isomer C-1 as well (Fig. 3c). The isomer populated first could not be accumulated strongly even at the low temperature but quickly reached a steady state concentration while the population of C-1 further increased upon continued irradiation. When switching off the light and raising the temperature slightly to − 125°C, the first formed isomer thermally converted exclusively into C-1 (Fig. 3d). The kinetic analysis delivered a Gibbs energy of activation of ΔG‡ = 9.67 kcal/mol for this process (see Supporting Information for details). This behavior is fully consistent with the predicted properties of isomer B-1, which could thus be assigned as the direct photoproduct of A-1. A complementary set of experiments was conducted to investigate the photoisomerization of isomer C-1. To our surprise irradiation of C-1 at − 105°C in THF-d8 solution resulted in the population of a fourth isomer, which could be accumulated almost quantitatively (Fig. 3e). This strong accumulation allowed a thorough 1H NMR analysis at low temperatures (–108°C to − 120°C) including NOE experiments to evidence the E configuration of the (supposed to be intact) double bond (see Supporting Information for all details of this analysis). However, upon closer scrutiny, 2D NMR analysis revealed that no carbonyl-carbon signal could be detected. Further, signal shifts of the expected central double bond-carbon atoms as well as the stereogenic carbon center also did not match with their expected hybridization or substitution-character. Thermal annealing of the new isomer at − 80°C in the dark led to full conversion to A-1 (Fig. 3f). The corresponding kinetic analysis revealed a Gibbs energy of activation of ΔG‡ = 13 kcal/mol for this process, which is in stark contrast to the expected barrier for thermal E-1 to A-1 conversion. Because of these accumulated experimental evidences and discrepancy to the calculations it became clear that this fourth isomer could not be E-1. However, remarkably good agreement between theoretically calculated and experimental NMR spectra were found for the epoxide constitutional isomer D-1 (see below). Moreover, also the calculated Gibbs energy of activation for thermal D-1 to E-1 conversion matches very well with the experimentally established value for the intermediate decay.
In order to fully support the tentative isomer assignment of the intermediate to D-1 as well as assignments of the other isomers, theoretically predicted NMR, UV/vis, and ECD spectra were compared to experimental ones (see Fig. 4 and the Supporting Information for more details). The theoretically predicted 1H NMR chemical shifts are in very good agreement with the experiments under the assumption that the photoproduct of C-1 irradiation is indeed D-1 (see Supporting Information for the detailed comparison of experimental and theoretical D-1 spectra). UV/vis spectra and ECD spectra of A-1, B-1, C-1, and D-1 could be directly compared between low temperature experiment and theory and again a very good agreement was found (Fig. 4a-d). To this end EPA (5:5:2 mixture of Et2O: i-pentane: EtOH) was used as solvent to access very low temperatures, which allowed to obtain the UV/vis and ECD spectra for the pure D-1 isomer at − 120°C and even the full spectral signature of the fleeting B-1 isomer at − 160°C. When irradiating enantiomerically pure A-1 at − 160°C the UV/vis spectrum displayed a bathochromic shift, while the ECD spectrum changed signs of the Cotton effect (Fig. 4e). This behavior reveals first, that the conjugation of the central double bond is not broken in the intermediate and second, that it possesses opposite helicity. Further the bathochromic absorption shift is indicative for a change from E to Z configuration of the central double bond similar to HTI photoswitches in general (note that in HTIs the E and Z nomenclature appears inverted because of the particular substitution pattern and resulting CIP priorities of 1). A very good match with the calculated UV/vis and ECD spectra of B-1 is observed in this case, which thus could confidently be assigned (Fig. 4b). Thermal annealing at − 108°C led to the known spectra of C-1 (Fig. 4e) directly reporting on full unidirectionality for the whole A-1 to B-1 to C-1 conversion sequence. When irradiating enantiomerically pure C-1 at − 120°C a very distinct behavior was observed (Fig. 4f). A new intermediate state formed with strongly hypsochromically shifted absorption. In fact, no absorption in the visible range remained after full conversion in the pss, which directly evidences breaking of the central double bond and thus loss of conjugation in the intermediate isomer. Besides the UV/vis absorption also the corresponding ECD spectrum could very well be matched with the theoretically predicted one of intermediate D-1 (Fig. 4d), which allowed us to now unambiguously assign this isomer. It was thus found that a highly unusual constitutional alteration takes place in the photochemical C-1 to D-1 transition, which retains a large amount of the incident lights energy according to the theoretical description. Thermal annealing at − 80°C led to full conversion to the A-1 isomer as evidenced by both UV/vis and ECD spectral changes without the observable formation of state E-1 (Fig. 4f). This behavior is expected from the very small calculated Gibbs energy of activation for the THI, which leads from E-1 to A-1. Therefore, D-1 seems to directly convert to A-1 at the temperature of the experiment but does in fact undergo first epoxide ring opening to E-1 and then quick follow-up THI to A-1, which cannot be evidenced individually by experiment.
Overall, the combined NMR, UV/vis and ECD experiments allowed an unambiguous isomer assignment confirming constitutional alteration as a key-step as well as complete unidirectionality of motor 1. Especially noteworthy is the completely selective conversion of D-1 to E-1 and then to A-1, since isomer D-1 inherits a single bond instead of the configurationally stable double bond as rotation axis. It is only because of the significant intramolecular hydrogen bond that the rotation direction is not reversed in the thermal follow up step, which would populate C-1 instead E-1. Our calculations show that D-1 is the global minimum of the epoxide state, which is not the case when the OH proton is replaced by a methyl group in silico (see Supporting Information for the corresponding data). The hydrogen bonding effect is present even in protic solvents like MeOD-d4 (see below and the Supporting Information) and despite an apparent proton-deuterium exchange. Thus, intramolecular hydrogen bonding is truly dictating the sense of directionality in the D-1 to A-1 rotation sequence and thus is responsible for motion control instead of sterics. Interestingly this is not the case for the A-1 to B-1 to C-1 rotation sequence, although intramolecular hydrogen bonding is present in isomer A-1 and needs to be broken in the photochemical step. However, we found that when the OH proton is replaced again by a methyl group in silico, the inherent directionality of the corresponding A-2 to B-2 to C-2 rotation sequence is retained and thus is the same as in motor 1. It becomes apparent that isomers A are more stable than isomers B within this molecular setup in general, which is a steric effect that favors the i-propyl group residing at the same side of the indanone plane as the phenyl-methoxy (see the Supporting Information for more details).
When comparing the kinetics for the thermal helix inversion from D-1 to A-1 (both intramolecularly hydrogen bonded) no strong influence of the solvent polarity was observed. Kinetic analysis in MeOH or THF delivered roughly the same Gibbs energy of activation of ΔG‡ = 13 kcal/mol for this process (see Supporting Information). It has to be emphasized however, that for this process no full disruption of the intramolecular hydrogen bond is needed for the conversion from D-1 to A-1 but rather a proton hopping during re-tautomerization and epoxide opening. Therefore, possible effects of the solvent are likely to cancel out rather than preferring one isomeric state and significantly altering the energy landscape.
It finally needs to be emphasized that a significant red shift of about 50 nm is seen for the absorptions of the stable A-1 and C-1 isomers when compared to the structurally related first HTI motor inheriting a sulfoxide as stereogenic center. Such red shift of absorption is desirable for many applications of molecular motors e.g. in the context of biology, catalysis, or materials and thus makes motor 1 a highly interesting candidate in this regard.
In conclusion, we describe a HTI-based light driven molecular motor 1 inheriting intramolecular hydrogen bonding. Different to all earlier HTI-motor setups, motor 1 receives its asymmetry from a carbon-based stereogenic center located at the indanone fragment. We demonstrate that this asymmetry is effectively translated into complete unidirectionality of the motor rotation. We further evidence that hydrogen bonding is in fact responsible for unidirectionality of this motor. A unique and distinct operation mechanism is established in which constitutional alteration and tautomerization processes allow to store an unprecedented large amount of the provided light energy within the motor rotation cycle. With this molecular setup a new type of molecular motors has become available that can be supercharged by light irradiation. This progress will open up an unexplored realm of motor applications where a significant energy budget or workload is expended e.g. in active mechanically driven processes45–50 or bulk material changes.51–55