Synthesis and Photoswitching of Ligand L

Photoresponsive subcomponent 1 was synthesised from 2-bromo-8-iodo-11,12-dihydrodibenzo[c,g][1,2]diazocine^[48] (Scheme S1 and Section S2) via two successive, selective, and high-yielding cross-coupling reactions with ethyl pinacol boronic acid esters^[49] 5-(4,4,5,5-tetraethyl-1,3,2-dioxaborolan-2-yl)-2,2′-bipyridine (S3) and 2-(1,3-dioxolan-2-yl)-5-(4,4,5,5-tetraethyl-1,3,2-dioxaborolan-2-yl)pyridine (S6).

At ambient conditions, subcomponent 1 primarily exists as its thermodynamically stable Z-isomer (99% Z-1). When exposed to 390 nm light, a second set of signals associated with E-1 emerged in the ¹H NMR spectrum, reaching a photostationary state of 72% E-1 and resulting in a colour change of the solution from yellow to red (Scheme 1, left; Section S7.4). E→Z isomerisation (>99% Z-1) took place under white light or 500 nm light irradiation and through thermal relaxation, exhibiting a half-life of 165 min at 25 °C in acetonitrile (20 µM; Section S7.3.1). The photoswitching of subcomponent 1 is rapid and completely reversible, usually taking less than a minute of irradiation in either direction (20 µM–2 mM in CH₃CN, 0.1–0.2 mW light power). The asymmetric 2,8-substitution pattern at the central diazocine unit in subcomponent 1 enables a significant structural change from a 90° to a 180° angle between the two metal-binding moieties upon photoswitching (Scheme 1, left; Figure S67). Employing two distinct metal-binding moieties generates directionality within asymmetric subcomponent 1. Dynamic covalent tethering of 1’s pyridine carboxaldehydes into TREN tri-pyridylimines fixes subcomponent 1’s orientation, leading to regioselective helicate self-assembly, as previously demonstrated^[26] in related systems.

Synthesis of Homometallic Helicates

Homometallic, triple-stranded helicates Fe₂L, Co₂L, and Zn₂L were obtained from the self-assembly of pyridine carboxaldehyde 1 (1 equiv.), tris(2-aminoethyl)amine (TREN, 1 equiv.), and the corresponding metal(II) trifluoromethanesulfonates (triflate or ⁻OTf, 2 equiv.), Fe(OTf)₂, Co(OTf)₂, and Zn(OTf)₂, respectively (Scheme 1, right). Fe₂L, Co₂L, and Zn₂L were characterised by mass spectrometry, NMR spectroscopy, and UV–vis spectroscopy (Sections S3 and S7.3).

Fe₂L crystallised in the triclinic space group P1 (Figure 1a; Section S6.2). Triple-stranded ligand L connects two homochiral metal centres, forming a dinuclear helicate structure with distorted C₃ symmetry. All Z-diazocine units maintain the same orientation within ligand L, with their helical configuration (L^MMM) opposing the helicity of the metal centres (Fe^ΔΔ, Figure 1a; Sections S6.1 and S6.2). The self-assembly of dinucelar homometallic helicate Fe₂L exhibits remarkable diastereoselectivity, resulting in the formation of only one pair of enantiomers: Fe^ΔFe^ΔL^MMM and Fe^ΛFe^ΛL^PPP.

The high-resolution electrospray ionisation (ESI) mass spectrum of Fe₂L showed a sequence of signals corresponding to the species [Fe₂L − nOTf]ⁿ⁺ (n = 1, 2, 3, and 4), confirming the stoichiometry of Fe₂L in the gas phase (Figures 2a and S24). ¹H NMR spectroscopy indicated a C₃ symmetric structure with diasterotopic signals for the TREN ethylene groups, as expected for the chiral coordination pocket (Figures 2b and S19). Proton signals of H-p, H-s, H-u′, and H-t′ close to the TREN tri-pyridylimine metal corner (Fe^TREN) are significantly broadened with half-widths of 30 to 50 Hz and are only visible in highly concentrated samples (Figure S19). ¹H,¹H ROESY spectroscopy revealed close interproton contacts between diazocine protons H-h and H-m with bipyridine proton H-g and pyridylimine proton H-p, respectively, confirming that the same racemic diastereomer of Fe₂L, namely Fe^ΔFe^ΔL^MMM, together with its enantiomer Fe^ΛFe^ΛL^PPP, is present in both solution and the solid state (Figure S56 and Section S6.4.1). The UV–vis spectrum of Fe₂L displays a diazocine n–π* absorption band at 405 nm, along with three bands in the visible range that are characteristic of Fe^II MLCT transitions (λ = 497, 539, and 599 nm; Figure 3a). By synthesising model compounds for the two unique metal binding sites in Fe₂L—Fe(bipy)₃(OTf)₂ and Fe(TREN tri-pyridylimine)(OTf)₂ (short: Fe(bipy)₃ and Fe(TRENpy), respectively)—and comparing their UV–vis spectra to that of Fe₂L (Figure 3b), we assigned the three MLCT transition bands to the two distinct coordination sites in Fe₂L. The two lower-wavelength bands are attributed to the MLCT transition of one Fe^II coordinated by three bipyridines (Fe^bipy, λ_bipy = 497 and 539 nm). The second Fe^II coordinated by the TREN tri-pyridylimine coordination pocket (Fe^TREN) causes the much weaker higher-wavelength band at λ_TREN = 599 nm. These findings align well with comparable Fe^II complexes documented in the literature.^{[50, 51]}

Characterisation of homometallic helicates Co₂L and Zn₂L by mass spectrometry and NMR spectroscopy indicated the formation of similar structures to Fe₂L (Figures 2 and 3; Sections S3.3, S3.4 and S6.4.2).

The ¹H NMR spectrum of Co₂L is characteristically dispersed over a range of nearly 200 ppm (Figure S26), due to the paramagnetism of Co^II, which precludes further assignments of the signals. The UV–vis spectrum of Co₂L shows an MLCT absorption band at λ ≈ 350 nm and an almost unchanged diazocine n–π* absorption band (405 nm; Figure S72).

The ¹H NMR spectrum of Zn₂L showed signal shifts similar to but clearly distinct from Fe₂L (Figure 2b). UV–vis spectroscopy of Zn₂L revealed an almost unchanged diazocine n–π* absorption band around 405 nm compared to subcomponent 1, with no MLCT transition observed in the visible range (Figures 3a and S70).

The characteristic proton shifts of H-h, H-m, H-p, H-t, and H-u of Fe₂L and Zn₂L in ¹H NMR (Figure 2b), as well as the three characteristic MLCT bands of Fe^bipy and Fe^TREN in UV–vis (Figure 3a), allow for detailed elucidation of Fe^II and Zn^II distribution between the two coordination sites of ligand L, enabling us to analyse heterobimetallic helicates precisely.

Synthesis of Heterobimetallic Helicates

Ligand L features two coordination sites that are chemically, kinetically, and thermodynamically distinct: a TREN tri-pyridylimine and a bipyridine. The sixfold chelation of the TREN tri-pyridylimine coordination pocket (M^TREN) is kinetically more inert and potentially thermodynamically preferred over the three twofold chelating bipyridines (M^bipy) due to an enhanced chelate effect similar to that of a cryptand. The greater stability of TREN tri-pyridylimine coordination sites, compared to three individual pyridylimines, has been used in previous applications for subcomponent exchange and cage stabilisation.^{[52, 53]} Considering kinetics, the bipyridine binding site (M^bipy) should be accessible first, as the TREN tris-pyridylimine coordination pocket (M^TREN) has not yet formed via metal-mediated imine-bond formation. Thus, employing ligand L alongside two different metal cations, with one forming thermodynamically more stable M─N bonds^[54] and exhibiting slower ligand exchange kinetics,^[55-57] should result in the selective assembly of heterobimetallic helicates. Potential combinations of metal cations that favour an octahedral coordination sphere include Fe^II and Zn^II or Co^II and Zn^II.

Using the greater stability and kinetic inertness of the TREN tri-pyridylimine binding pocket (M^TREN), we attempted the consecutive assembly of a heterobimetallic helicate by mixing pyridine carboxaldehyde 1 (3 equiv.), TREN (1 equiv.), and Fe(OTf)₂ (1 equiv.) to form mononuclear complex FeL with Fe^II in the TREN tris-pyridylimine binding pocket (Fe^TRENL, Figure 4a). Subsequently, Zn(OTf)₂ (1 equiv.) was added to FeL to allow formation of a heterobimetallic helicate with Fe^II remaining in the TREN tris-pyridylimine binding pocket (Fe^TREN) and Zn^II coordinating to the three bipyridine units (Zn^bipy), which we refer to as FeZnL (short for Fe^TRENZn^bipyL). High-resolution ESI mass spectrometry confirmed the formation of an [Fe + L] species following the first reaction step, with minor amounts of Fe₂L also detected. After the second reaction step, an [Fe + Zn + L] species was generated, with some Fe₂L still present alongside newly formed Zn₂L (Figures 2a and S31). The presence of both Fe₂L and Zn₂L, alongside the desired FeZnL, suggests that the formation of Fe₂L is thermodynamically more favourable than the formation of FeL, which results in unreacted or partially reacted TREN and 1 remaining in the reaction mixture. After the addition of Zn(OTf)₂, the unreacted or partially reacted ligand subcomponents, TREN and 1, form Zn₂L.

The ¹H NMR spectrum of FeZnL confirmed the formation of a heterobimetallic helicate after the second reaction step (Figures 2b and S30). The chemical shifts of protons near the TREN tris-pyridylimine binding pocket, specifically diazocine proton H-m, pyridylimine proton H-p, and ethylene protons H-t and H-u, are comparable to the chemical shifts of their Fe₂L counterparts. In contrast, the chemical shift of diazocine proton H-h, which is close to the bipy unit, is nearly identical to that of its Zn₂L counterpart, indicating an Fe^TRENZn^bipyL structure. As indicated by the ESI mass spectrum, ¹H NMR spectroscopy also suggested the formation of dinuclear homometallic Fe₂L and Zn₂L (25% and 10%, respectively, Figure 2b; Section S3.7). The higher proportion of Fe₂L compared to Zn₂L is likely due to an excess of Fe(OTf)₂ (approximately 10%) present during the formation of FeL. As described above, distinct MLCT absorption bands for Fe^II, coordinated by three bipyridine units (Fe^bipy) or a TREN tris-pyridylimine binding pocket (Fe^TREN), facilitated further structural conformation of FeL and FeZnL using UV-vis spectroscopy (Figures 3 and S32). The UV–vis spectra of Fe^TRENL and the reference compound Fe(TRENpy) are nearly identical, suggesting the presence of an Fe^TREN coordination sphere in Fe^TRENL. The UV–vis spectrum of FeZnL reveals a secondary band associated with the additional MLCT transition observed in the contaminant Fe₂L. Still, it displays a major band that is identical to the one found in Fe^TRENL. The difference in absorption bands between Fe^bipy and Fe^TREN is readily visible to the naked eye, as the bipyridine-containing Fe^II complexes exhibit a vibrant red colour, whereas their pyridylimine counterparts present a deep purple hue (Figure 3, insert). The kinetics of Fe^TRENL complex formation were further investigated using UV–vis spectroscopy (Section S5), which indicated the nearly immediate formation of an Fe^bipy1₃ complex, in which Fe^II is coordinated by three bipy units, as evidenced by the appearance of the characteristic Fe^bipy-MLCT absorption band. Over a few hours, the Fe^bipy-MLCT absorption band transformed into an Fe^TREN-MLCT absorption band, thus indicating the successful formation of Fe^TRENL. The progress of the reaction can also be observed with the naked eye: the reaction solution turns a distinctive Fe^bipy-red colour within minutes and then gradually changes to a deep violet shade, characteristic of Fe^TREN, indicating the sequential formation of Fe^bipy1₃ and Fe^TRENL. Therefore, we conclude that Fe^bipy1₃ is kinetically favoured, whereas Fe^TRENL is potentially thermodynamically more stable but also kinetically more inert, thus confirming our assumption that the TREN tris-pyridylimine coordination pocket should be the more stable binding site.

After confirming our assumption regarding the kinetic and thermodynamic coordination behaviours of aldehyde 1 and ligand L, we aimed to harness these properties to achieve heterobimetallic social self-sorting using Fe^II, Zn^II, and Co^II. Given that Fe^II and Zn^II show the most significant difference in thermodynamic and kinetic stabilities (Figure 4b), we initially sought to form a self-sorted ZnFeL heterobimetallic helicate (short for Zn^TRENFe^bipyL) from a mixture of pyridine carboxaldehyde 1 (3 equiv.), TREN (1 equiv.), Fe(OTf)₂ (1 equiv.), and Zn(OTf)₂ (1 equiv.) in acetonitrile (Figure 4a). The self-assembly of Fe^TRENL indicated that the M^bipy coordination site would form before the M^TREN binding site because it is readily accessible. In contrast, the M^TREN coordination site must be formed through three metal-mediated imine condensation reactions, which occur on a significantly slower timescale. Consequently, we anticipated that both Fe^II and Zn^II would first coordinate to the bipy units of 1, with Fe^II gradually replacing the thermodynamically less stable Zn^II (ΔΔG^Fe,Zn(M(bipy)₃²⁺) = 52.2 kJ mol⁻¹)^[54] at all coordination sites, resulting in the formation of Fe^bipy1₃. Once all Fe^II is coordinated to bipy (Fe^bipy), only Zn^II should remain to facilitate the formation of TREN tri-pyridylimine, thereby selectively installing Zn^II in the M^TREN coordination pocket to produce ZnFeL (short for Zn^TRENFe^bipyL). ZnFeL crystallised in the triclinic space group P1 (Figure 1b; Section S6.2). The solid-state structure confirmed the expected Zn^TREN-Fe^bipy metal distribution in a triple-stranded helicate structure almost identical to that of Fe₂L (Figure 1a; Section S6.2). High-resolution ESI mass spectrometry confirmed the formation of a [Zn + Fe + L] heterobimetallic helicate (Figures 2a and S38). ¹H NMR spectroscopy indicated the expected regioselectivity of the self-sorted helicate ZnFeL (Zn^TRENFe^bipyL), as evidenced by the characteristic proton signal shifts of protons H-h, H-m, H-p, H-t, and H-u corresponding to Zn^TREN and Fe^bipy coordination (Figure 2b). Furthermore, ¹H,¹H ROESY NMR spectroscopy confirmed the retention of the solid-state structure in solution (Section S6.4.1). UV–vis spectroscopy revealed a characteristic Fe^bipy-MLCT absorption band (λ_bipy = 537 nm), whereas no Fe^TREN-MLCT absorption band (λ_TREN ≈ 580 nm) was detected, further supporting the successful self-sorting within ZnFeL (Figures 3b and S76).

The investigation of the kinetics of ZnFeL complex formation via UV–vis spectroscopy further corroborated our proposed assembly mechanism for ZnFeL (Section S5). An Fe^bipy-MLCT absorption band (λ_bipy = 540 nm) quickly appeared, with no further changes to the spectra, indicating that Fe^II remained in the bipy binding site, thereby enabling the selective formation of TREN tri-pyridylimine around Zn^II. These findings suggest that this self-sorting into ZnFeL is, indeed, a delicate interplay between ligand association and formation kinetics as well as metal-coordination thermodynamics. Prolonged heating of ZnFeL (65 °C, 2 months) in the presence of 3.0 equiv. Fe^II led to the formation of 6% Fe₂L, demonstrating the kinetic inertness of the ZnFeL structure and the Zn^TREN coordination site while indicating that Fe₂L might be thermodynamically favoured over ZnFeL (Section S9.4).

To support our hypothesis that heterobimetallic helicate self-sorting occurs due to a delicate interplay between metal–ligand stability and exchange kinetics, we attempted the synthesis of two additional heterobimetallic helicates using Zn^II/Co^II and Co^II/Fe^II mixtures. In these mixtures, both the differences in thermodynamic metal–ligand stability and in metal–ligand exchange kinetics are gradually reduced. While the difference in metal–ligand exchange kinetics between Zn^II and Co^II is comparable to that of the Zn^II/Fe^II pair,^[55-57] the difference in M^bipy-association energy between Co(bipy)₃²⁺ and Zn(bipy)₃²⁺ is approximately half as high as that between Fe(bipy)₃²⁺ and Zn(bipy)₃²⁺ (ΔΔG^Fe,Zn(M(bipy)₃²⁺) = 52.2 kJ mol⁻¹ and ΔΔG^Co,Zn(M(bipy)₃²⁺) = 28.0 kJ mol⁻¹, respectively; Figure 4b).^[54] Therefore, helicate self-assembly from a mixture of Zn^II and Co^II should indicate whether the difference in thermodynamics or kinetics of the metal cations has a more significant impact on successful self-sorting. For Co^II and Fe^II, both the difference in M^bipy-association energy between Co(bipy)₃²⁺ and Fe(bipy)₃²⁺ and the difference in metal–ligand exchange kinetics are small.^[54-57] Consequently, if our hypothesis was correct, we expected poor self-sorting, if any.

In the case of the one-pot heterobimetallic helicate assembly involving Zn^II, Co^II, 1, and TREN (Figure 4a), both ¹H NMR spectroscopy and ESI mass spectrometry indicated selective self-sorting into the ZnCoL (Zn^TRENCo^bipyL) helicate (SI,Section S4.1.2). ESI mass spectrometry revealed a singular set of peaks for [Zn + Co + L + nOTf]⁽⁴⁻ⁿ⁾⁺ (n = 0, 1, 2), and the wide-sweep ¹H NMR spectrum displayed only signals reminiscent of those in Co(bipy)₃(OTf)₂, with none reminiscent of a Co^II-TREN tri-pyridylimine coordination.

Helicate assembly in the presence of Co^II and Fe^II, however, resulted in a mixture of CoFeL (Co^TRENFe^bipyL, approx. 60%–70%), Fe₂L (approx. 15%) and Co₂L helicates (approx. 15%). The composition of the helicate mixture was tentatively estimated using ESI mass spectrometry, NMR, and UV–vis spectroscopy (Sections S4.3 and S4.4). The paramagnetic nature of Co^II hindered the determination of accurate values.

Photoswitching of Zn₂L

After demonstrating the ability of ligand L to form either dinuclear homometallic helicates or even heterobimetallic ones via the effective self-sorting of two different metal cations, we investigated the effects of switching the photochromic diazocine units within the helicates.

Irradiation of an acetonitrile solution of Zn₂L with visible light (405 nm, 25 °C, 1 mM) resulted in a colour change from yellow to red, along with the disappearance of the 405 nm absorption band and the emergence of an intense absorption band at 490 nm in the UV–vis spectrum (Figure 5a). The changes in the UV–vis spectrum are nearly identical to those observed for aldehyde 1 (Figure S80), suggesting an equally efficient Z→E conversion in Zn₂L helicate as in free aldehyde 1 upon irradiation with 405 nm light. High-resolution ESI mass spectrometry demonstrates that the Zn₂L composition remains intact following irradiation with 405 nm light (Figure S99). Additionally, ¹H DOSY NMR spectroscopy indicated that the diffusion coefficients—and, therefore, the solvodynamic diameters—of the structures before and after 405 nm light irradiation differ only slightly (Figure 6b, bottom). In the ¹H NMR spectrum, irradiation of Zn₂L with 405 nm light results in the disappearance of nearly all previously sharp and well-defined signals, leaving only ill-defined and broad signals, apart from the signals of the ethylene groups H-t and H-u. Moreover, no signals for the free ligand L or free aldehyde 1 were detected, indicating that at least the Zn^TREN binding pocket remained intact following irradiation (Figure 6b, green, 2^nd from top). The intact Zn^TREN binding pocket, along with high-resolution ESI mass spectrometry confirming the retention of the Zn₂L composition, ¹H DOSY NMR spectroscopy suggesting a structure of a similar solvodynamic diameter, and UV–vis spectroscopy demonstrating successful switching of the diazocine units, might indicate the formation of an E,E,E-Zn₂L structure (abbreviated as E-Zn₂L, Figure 6a). Although switching should occur in a stepwise manner, we tentatively assume structures in which only some of the diazocines are switched, namely E,E,Z-Zn₂L and E,Z,Z-Zn₂L, to be unlikely, as literature precedence^{[14, 18, 58, 59]} of cages containing photochromic units in their ligand backbone suggests that switching is a highly cooperative process, with intermediates only observed at extremely low temperatures.^[60] The first switching event causes the most significant structural change, potentially accelerating the subsequent switching of other photochromic units within the same structure to relieve strains caused by previous isomerisation events.

The observed broad ¹H NMR signals after 405 nm irradiation of Zn₂L (Figure 6b, green, 2^nd from top) indicate structural transformations, which are on intermediate exchange on the NMR timescale. We tentatively assume these to be hindered rotations of the diazocine moieties within the arms of the ligands or isomerisation of the helicate to a pseudo-mesocate. In E,E,E-Zn₂L, the three “arms” of ligand L adopt a pseudo-linear configuration, enabling rotation of the diazocine moieties (i-E-Zn₂L, Figure 6a, right; Figure S102).

Rotation of the diazocine moieties would give rise to four potential constantly interconverting rotamers with the diazocine units pointing with their azo-nitrogens either in a clockwise (L^PPP) or counterclockwise direction (L^MMM) in the helicate or even mixtures of both (L^PPM or L^MMP; Figure 6a). Additionally, due to steric strains within E-Zn₂L, we anticipate that both Zn^II corners in E-Zn₂L will be significantly weaker than those in Z-Zn₂L, potentially promoting isomerisation between helicate and pseudo-mesocate or even facilitating metal exchange.

To test our hypothesis regarding the interconversion of rotamers and helicate-to-pseudo-mesocate leading to the broad signals in the ¹H NMR spectrum after 405 nm irradiation of Z-Zn₂L, we performed low-temperature ¹H NMR spectroscopy (240 K) on the same sample (Figure 6b, orange, 3^rd from top; Figure S97). The resulting spectrum reveals a multitude of sharp signals, indicating a plethora of chemical environments and isomeric structures, which do not interconvert at this temperature. This observation supports our hypothesis that multiple interconverting diastereomers (i-E-Zn₂L) are responsible for the broad signals observed at 298 K. Irradiation of i-E-Zn₂L with either 500 nm or white light reverses the process almost instantly, indicating that Zn₂L switching is completely reversible at ambient temperature (Figure 6b, black, 4^th from top). However, at 240 K, irradiation of i-E-Zn₂L with white light does not lead to the clean formation of Z-Zn₂L but rather a multitude of sharp signals that indicate the formation of multiple isomers (Figure S97). Therefore, we tentatively assume that irradiation with white light causes all diazocine moieties to switch back into their Z states, with each i-E-Zn₂L helicate or pseudo-mesocate rotamer isomerising to its corresponding i-Z-Zn₂L helicate or pseudo-mesocate atropisomer. These atropisomers cannot interconvert without breaking at least one pair of Zn^II-bipy coordinative bonds. At 240 K, the abundance of atropisomers results in numerous sharp ¹H NMR signals (Figure S97). At room temperature, the Zn^II-bipy coordinative bond readily dissociates, allowing the atropisomers to converge to the most stable Z-Zn₂L structure (Figure S97). Thus, the isomerisation process Z-Zn₂L→E-Zn₂L→Z-Zn₂L is influenced by the kinetic lability of the metal cation within the bipy pocket of L. To further explore the intricate Z-Zn₂L→E-Zn₂L→Z-Zn₂L switching mechanism and support our hypothesis, we investigated switching in ZnFeL, which contains a kinetically more stable Fe^II in the bipy binding site of L (Fe^bipy) and should, thus, hinder isomerisation of the atropisomers (i-Z-ZnFeL) to the thermodynamically preferred structure (Z-Zn^ΔFe^ΔL^MMM).

Photoswitching of ZnFeL: A Molecular Energy Ratchet

Similar to Zn₂L, the irradiation of ZnFeL with 405 nm light resulted in a ¹H NMR spectrum featuring broad signals, while ¹H DOSY NMR spectroscopy and high-resolution ESI mass spectrometry indicated the formation of a species with the same composition and a similar size to Z-ZnFeL, which we denote as i-E-ZnFeL (Figure 6; Sections S7.4 and S7.5). The UV–vis spectra of Z-ZnFeL and E-ZnFeL showed nearly the same differences as those between Z-Zn₂L and E-Zn₂L, as well as Z-1 and E-1, with the MLCT absorption band remaining unchanged (Figure 5b; Figure S80). This suggests that photoswitching remains efficient, even in the presence of the Fe^bipy MLCT absorption band, and that both Fe^TREN and Fe^bipy coordination are preserved. As with i-E-Zn₂L, the observed broad ¹H NMR signals for i-E-ZnFeL sharpened when cooled to 240 K (Figure 6c, orange, 3^rd from top), indicating that diazocine rotation and helicate→pseudo-mesocate isomerisation caused the broadened signals at room temperature.

As envisioned, irradiating i-E-ZnFeL with white light resulted in a complex ¹H NMR spectrum featuring multiple signal sets (Figure 6c, blue, 4^th from top). This signal pattern is similar to that of i-E-Zn₂L following 405 nm and white light irradiation at 240 K, prompting us to label it i-Z-ZnFeL (Scheme 2; Figure S98). We can reform the thermodynamically stable Z-ZnFeL by heating i-Z-ZnFeL at 65 °C for five hours (Figure 6c, black, 5^th from top). The formation of kinetically trapped states after 405 nm and white light irradiation at room temperature was also observed for Fe₂L (Figure S84), supporting our hypothesis that the dissociation of bipyridine units in the M^bipy coordination site is the rate-determining step for the kinetically trapped states to revert to the thermodynamically favoured Z-M^ΔM^ΔL^MMM (Z-Zn^ΔZn^ΔL^MMM, Z-Zn^ΔFe^ΔL^MMM, and Z-Fe^ΔFe^ΔL^MMM, respectively). The stronger metal–ligand bond of Fe^II compared to Zn^II leads to a higher barrier for the dissociation of metal–bipyridine coordination, creating a more effective kinetic trap. This elevated barrier is evident in the temperature needed for the various helicate and pseudo-mesocate atropisomers (i-Z-Zn₂L, i-Z-ZnFeL, and i-Z-Fe₂L) to return to their initial thermodynamically favoured states Z-M^ΔM^ΔL^MMM (Z-Zn^ΔZn^ΔL^MMM, Z-Zn^ΔFe^ΔL^MMM, and Z-Fe^ΔFe^ΔL^MMM, respectively), which rises from room temperature for i-Z-Zn₂L to 65 °C for i-Z-ZnFeL and i-Z-Fe₂L.

We examined the kinetically trapped intermediates during the switching cycle of ZnFeL through in situ illuminated ¹H NMR spectroscopy (Section S7.4.3). Although the variety of signals hindered a thorough kinetic analysis, our findings indicate that irradiating ZnFeL with 405 nm light produces a mixture of at least two distinct species. This implies that the structural alterations induced by 405 nm light irradiation result in a distribution of conformational and possibly isomeric (helicate and pseudo-mesocate) states. The notable asymmetry in these i-E-ZnFeL states is evident in the intricate ¹H NMR spectrum.

Trapping the respective i-E-ZnFeL in their current isomeric forms through photoswitching provides a potential direct pathway from symmetric E-Zn^ΔFe^ΔL^MMM to ground state Z-Zn^ΔFe^ΔL^MMM (Scheme 2, process V).

In the case of Zn₂L, i-Z-Zn₂L are kinetically trapped at low temperatures (Figure 6b), demonstrating that the same processes operate during the irradiation of Zn₂L.

Since most molecules follow the reaction cycle unidirectionally (Scheme 2, processes I–IV), the photoswitching of ZnFeL acts as a light-driven molecular energy ratchet: 405 nm light irradiation drives the thermodynamically stable Z-ZnFeL helicate into metastable E-ZnFeL helicates, which can readily access various conformers and stereoisomers (i-E-ZnFeL). Under white light irradiation, these i-E-ZnFeL states relax swiftly into their corresponding kinetically trapped i-Z-ZnFeL isomers. These higher-energy kinetically trapped isomers (i-Z-ZnFeL) cannot readily return to the initial thermodynamically stable state (Z-ZnFeL) due to the required rotation of the diazocine units being only possible through partial Fe^bipy coordination-site dissociation. Thus, the system captures light energy to convert the more stable isomer (Z-ZnFeL) into less stable, kinetically trapped isomers (i-Z-ZnFeL). This transformation leads to the occupancy of higher-energy states, storing some of the absorbed light energy as strained or unfavourable molecular conformations.

An Autonomously Operating Molecular Ratchet Under White Light Irradiation

Kinetic studies on the thermal relaxation of the i-Z-ZnFeL mixture, obtained via subsequent 405 nm and white light irradiation, support the thermal relaxation pathway (ii). The time-dependent ¹H NMR spectra revealed that various components of this mixture were interconverting, with none of the involved species returning to the initial ground state of Z-ZnFeL (Z-Zn^ΔFe^ΔL^MMM; Section S7.4.4). Thermal conversion of pseudo-mesocate Z-Zn^ΛFe^ΔL^MMM into the overall thermodynamically more stable Z-Zn^ΔFe^ΔL^MMM helicate occurred at room temperature, exhibiting a half-life of several days (Figure S115).

Curiously, the molecular ratchet functions in sunlight as well. When a sample of Z-Zn^ΔFe^ΔL^MMM helicate is placed by a window for one winter day, it transforms into a mixture enriched with pseudo-mesocate Z-Zn^ΛFe^ΔL^MMM (Figure 7b, bottom). However, this process is less efficient compared to using a white LED. Therefore, the molecular ratchet is capable of converting light into chemical energy and can even retain that energy for a limited duration.

Metal Exchange

As mentioned earlier, ligand L contains two coordination sites with distinct kinetic stabilities, which we have shown to be essential for self-sorting of heterobimetallic helicates. To take advantage of the differences in kinetic inertness between the two binding sites, we performed a selective metal-cation exchange in the M^bipy-coordination site, transforming Zn₂L into ZnFeL via Zn^II→Fe^II exchange (Figure 8, top). Specifically, treating an acetonitrile solution of Zn₂L with Fe(OTf)₂ (1.5 equiv.) at 65 °C led to the formation of ZnFeL, confirmed by ¹H NMR, UV-vis spectroscopy, and ESI mass spectrometry (Section S9). Kinetic studies using ¹H NMR spectroscopy (Section S9.3) indicated that the Zn^II→Fe^II exchange does not follow apparent first-order kinetics. Rather, Zn₂L is rapidly consumed, resulting in both the intended product ZnFeL and a non-productive intermediate. This resting state is subsequently converted into the product ZnFeL via Zn₂L as an intermediate.

Given that the helicate molecular ratchet is accumulating high-energy structures (i-E-ZnFeL), in which the M^bipy coordination site is likely weakened, we aimed to examine whether we could use this unidirectional reaction cycle (Scheme 2) to expedite the Zn^II→Fe^II cation exchange within Zn₂L, resulting in the formation of ZnFeL. To evaluate this hypothesis, Fe(OTf)₂ (1.5 equiv.) was added to an acetonitrile solution of Zn₂L at 65 °C, and the mixture was illuminated with 405 nm light for 60 seconds every 25 min. Since ZnFeL shows a distinct Fe^bipy MLCT transition band at 550 nm, which correlates with the number of Fe─N bonds formed and which Zn₂L entirely lacks (Figure 5), this reaction can also be monitored in situ using UV–vis spectroscopy (Figure 8). Note that E-diazocines exhibit strong absorption at 550 nm (Figure S68), and this will influence the absorption band at 550 nm during irradiation, continuing until the diazocine thermally returns to its Z-form. The thermal relaxation of the E-diazocine moieties to their Z-configurations occurs rapidly at 65 °C (τ_½(65 °C) = 72 s), achieving complete conversion within less than 10 min. In the dark, the Zn₂L→ZnFeL background reaction (Figure 8, blue dotted line) shows a rapid initial rise in conversion before stabilising to a nearly linear reaction rate after about 12 min. Irradiating the Zn₂L→ZnFeL reaction mixture with 405 nm light outside the UV–vis spectrometer results in a significant increase in absorption at 550 nm (Figure 8, green dashed line), primarily attributed to the absorption of the E-diazocine units. However, the observed increase in 550 nm absorption approximately 10 min after irradiation can be solely attributed to an increased Fe^bipy absorption, which reflects the Zn₂L→ZnFeL conversion. Four subsequent irradiations of the reaction mixture with 405 nm light further accelerate the conversion of Zn₂L to ZnFeL, resulting in nearly double the amount of ZnFeL produced compared to the background reaction after 150 min. The increased conversion of Zn^bipy to Fe^bipy after five 60-s light irradiations of the ZnFeL-Fe(OTf)₂ reaction mixture reinforces our hypothesis that the molecular ratchet can help promote this exchange.

Notably, the Zn^bipy→Fe^bipy exchange could also be accelerated under constant irradiation (60 min; Figure 8, orange solid line). Once the E-diazocine moieties have reached their photostationary states, their absorption remains constant under continuous irradiation. Consequently, the significant fivefold steeper increase in absorption observed between minutes 10 and 60, compared to the background reaction, can be solely ascribed to an accelerated exchange from Zn^bipy to Fe^bipy.

We believe that the molecular energy ratchet mechanism is vital for enabling the metal-cation exchange process. By populating higher-energy states and enriching non-ideal conformers, the photoswitching process effectively destabilises the coordination sphere around the metal centres, thereby accelerating the Zn^bipy→Fe^bipy exchange process. The ability to control this process through light irradiation offers a powerful tool for modulating kinetics in complex systems.