Multistate Switching of Spin Selectivity in Electron Transport through Light‐Driven Molecular Motors

Abstract It is established that electron transmission through chiral molecules depends on the electron's spin. This phenomenon, termed the chiral‐induced spin selectivity (CISS), effect has been observed in chiral molecules, supramolecular structures, polymers, and metal‐organic films. Which spin is preferred in the transmission depends on the handedness of the system and the tunneling direction of the electrons. Molecular motors based on overcrowded alkenes show multiple inversions of helical chirality under light irradiation and thermal relaxation. The authors found here multistate switching of spin selectivity in electron transfer through first generation molecular motors based on the four accessible distinct helical configurations, measured by magnetic‐conductive atomic force microscopy. It is shown that the helical state dictates the molecular organization on the surface. The efficient spin polarization observed in the photostationary state of the right‐handed motor coupled with the modulation of spin selectivity through the controlled sequence of helical states, opens opportunities to tune spin selectivity on‐demand with high spatio‐temporal precision. An energetic analysis correlates the spin injection barrier with the extent of spin polarization.

An aqueous solution of LiOH aq (2.0 M, 0.35 mL, 10.0 equiv.) was added to mixture of (±)-E-2 (40.0 mg, 0.066 mmol, 1.0 equiv.) in THF (2 mL) and MeOH (2 mL) and the reaction mixture was heated at 50 C for 2 h. Subsequently the mixture was concentrated in vacuo, dissolved in water and quenched with aqueous HCl (1 M, 2.5 mL). Next, the mixture was extracted with EtOAc (3 x 20 mL) and the organic layer was washed with water (5 x 20 mL) and brine (2 x 20 mL), dried over MgSO 4 and subsequently the solvent was removed under reduced pressure. The resulting pale yellow solid was triturated with pentane/CH 2 Cl 2 (15/1 v/v) and dried in vacuo to yield (±)-E-1 as white solid (36 mg, 0.062 mmol, 94 %

Analysis of the rotary motion of molecular motor in solution
The rotary motion of molecular motor E/Z-1 was studied in solution using 1 H NMR, UV/Vis absorption and CD spectroscopy (Scheme S3a, Figures S18-24). Photochemical isomerization steps E-1-stable→Z-1-metastable (Process 1, Error! Reference source not found.a, Figure S22) and Z-1-stable → E-1-metastable (Process 3, Error! Reference source not found.a, Figure S23) were studied with 1 H NMR spectroscopy to determine the ratios of the respective metastable and stable isomers at the photostationary state (PSS) established upon irradiation at 312 nm. For E-1stable → Z-1-metastable photoisomerization, a characteristic downfield shift of the resonances assigned to the aliphatic protons were observed in line with the formation of the strained diastereomer. Photoisomerization from Z-1-stable to E-1-metastable shifted most of the resonances of the aliphatic protons downfield. Conversely, an upfield shift of the resonance corresponding to the methyl group adjacent to the stereogenic center was observed, characteristic for the Emetastable diastereomer of 1 st generation molecular motors. Heating of the irradiated samples led to a clean conversion of the metastable isomers to stable isomers, which could be readily identified and characterized with 1 H NMR spectroscopy ( Figure S22, Figure S23 blue spectra) Accordingly, for both of these UV-induced processes a large bathochromic shift of the main absorption band was observed in the UV/Vis absorption spectra consistent with the formation of the metastable isomers (Figures Figure S18a and Figure S20a). Furthermore, isosbestic points at 257 nm for Z-1-stable → E-1-metastable ( Figure S20) and at 263 nm and 330 nm for E-1stable → Z-1-metastable were maintained throughout both photochemical isomerizations (Processes 1 and 3, Error! Reference source not found.a) indicating clean unimolecular processes. For both E-1-stable → Z-1-metastable and Z-1-stable → E-1-metastable processes, CD spectroscopy revealed inversion of the CD signal, corresponding to the lowest-energy transition, upon irradiation at 312 nm (Figure S18b,c and Figure S20b,c). This is in line with the formation of the isomers with opposite helical chirality. The thermal steps (Process 2 and 4, Error! Reference source not found.a) were also studied by UV/Vis absorption spectroscopy. Upon thermal helix inversion, the metastable isomers were cleanly converted to their respective stable isomer. During this process, isosbestic points at 251 nm and 233 nm E-1-metastable→E-1-stable ( Figure S19a) and at 267 nm and 335 nm Z-1-metastable→Z-1-stable ( Figure S21a) were maintained. CD spectroscopy showed an inversion of signal during photoisomerization, in agreement with the inversion of the helical chirality of the motors upon conversion to the respective metastable isomers (Figure S18b  Figure S21d,c). A value of Δ ‡ G (20 °C) = 100.1 ± 0.4 kJ mol -1 (t 1/2 = 21 h) was found for the THI from Z-1-metastable→Z-1-stable. As expected, the THI of E-1-metastable to E-1stable was found to be fast at room temperature, Δ ‡ G (20 °C) = 78 ± 6 kJ mol -1 (t 1/2 = 10 s). Finally, irradiation of the sample of stable Z-1 isomer at 312 nm at room temperature led to essentially same photostationary state mixture (~27:73 ratio of stable E-1 to metastable Z-1 isomers as inferred from integration of 1 H NMR resonances) as the photostationary state mixture obtained starting from stable E-1 isomer ( Figure S22, Figure S24), thus demonstrating that in solution these motors operate as three switches at ambient conditions (Error! Reference source not found.b).
In conclusion, these measurements demonstrated that the photochemical and thermal isomerization behavior of this overcrowded-olefin molecular motor scaffold are only marginally affected by the S19 substitution used in this study. Based on this data, it can thus be concluded that molecular motor E/Z-1 behaves like the structurally related parent diol motor 3. [1] Scheme S3.

Raman Studies in solution
In addition, isomerization cycle of motor 1 was followed in solution by Raman spectroscopy. It was found that the respective diastereoisomers could be readily identified by Raman shifts of the characteristic bands. Irradiation of stable E-1 solution at 300 nm to gradual decrease in intensity centered at 1626 cm -1 ascribed to the aromatic ring mode of the stable isomer and gradual appearance of a new band centered at 1565 cm -1 , in line with the formation of the metastable Z-1 ( Figure S25a). Heating of this sample resulted in a gradual decrease in the intensity of the band centered at 1565 cm -1 accompanied by the increase in the intensity of the band at 1636 cm -1 ( Figure  S25b). The shift of the band characteristic of the stable isomer from 1626 cm -1 to 1636 cm -1 is consistent with formation of the stable Z-1 isomer from the metastable isomer and demonstrates that these species can be clearly distinguished with Raman spectroscopy ( Figure S25c). Consequently, irradiation at 300 nm of the solution of stable Z-1 at -50 ºC led to gradual decrease in the intensity of the band centered 1638 cm -1 and appearance of the new band at 1573 cm -1 ( Figure S26a) while heating of this sample resulted in the formation of new band at 1629 cm -1 ( Figure S26b) consistent with the sequential stable Z-1 to metastable E-1 to stable E-1 isomerization photochemical and thermal isomerization sequence ( Figure S26c). Finally, irradiation of the stable Z-1 solution at RT resulted in a gradual decrease of the band characteristic of stable Z-1 (1638 cm -1 ) with concomitant appearance of the band characteristic of the stable E-1 isomer (1626 cm -1 ) and finally formation of the band characteristic of metastable Z-1 isomer (1565 cm -1 ) in line with the formation of the metastable Z-1 via unidirectional sequence of photochemical and thermal isomerization reactions typical of 1 st generation molecular motors ( Figure S27d, Error! Reference source not found.a,b).

Raman studies in the solid state
Raman micropectroscopy was proved to be a suitable tool to follow the photochemical and thermal isomerization of molecular motors E/Z-1 in the solid state. In line with the studies performed in the solution, both powdered samples of E/Z-1 showed significant differences in the Raman shifts of the bands that were ascribed to aromatic ring mode and overcrowded double bond stretching ( Figure  S28a,b). In addition, several other characteristic bands were found in the fingerprint region of Raman spectra of both E/Z-1 motors, allowing for unambiguous identification of either of the diasteroisomers. Irradiation of the drop casted sample of Z-1 ( Figure S29a) at 300 nm at RT led to the appearance of a new band in the Raman spectrum centered at 1572 cm -1 (Figure S30a,b) in line with the formation of the metastable isomer of the molecular motor 1. Comparison to the data collected in the solution (1573 cm -1 and 1565 cm -1 for metastable E-1 and Z-1, respectively) allowed to ascribe this band to the metastable E-1 isomer. Furthermore, the intensity of this band gradually decreased ( Figure S31a) over time, in line with the thermal helix inversion step, ultimately yielding a spectrum consistent with stable E-1 isomer ( Figure S30a,b,c), thus further supporting formation of the metastable E-1 isomer. The half-life of the metastable E-1 was calculated by following the gradual decrease in the intensity of the band centered at 1571 cm -1 and was found to be remarkably long (ca. t 1/2 = 50 min.) in comparison to the solution (t 1/2 = 10 s) even under local-heating caused by the direct exposure to 785 nm 500 mW Raman laser ( Figure S31b). It should be noted that the half-life of metastable E-1 in solid is expected to be significantly longer for the sample kept in the dark (without exposure the 785 nm laser). Accordingly, irradiation of the drop casted E-1 sample ( Figure S29b) at 300 nm led to appearance of a new band in the Raman spectrum centered at 1562 cm -1 (Figure S32a,b) characteristic of the metastable Z-1 (in solution 1573 cm -1 and 1565 cm -1 for metastable E-1 and Z-1, respectively). However, in contrast to E-1, exposure of Z-1 to 300 nm light led to decomposition of the sample which was manifested as a large decrease in the overall scattering intensity in the Raman spectrum upon. Therefore, in order to avoid the complete decomposition, samples were irradiated for 10 min. and consequently the conversion to the metastable Z-1 was low ( Figure S32a,b). Furthermore, attempts to mitigate this decomposition by exclusion of oxygen, that is preparation of the sample in a glovebox and subsequent transport and measurement in a sealed quartz cuvette failed, thus indicating that the most probable decomposition pathway is photopolymerization of either stable E-1 or metastable Z-1 under 300 nm UV light. It should be emphasized however, that the Raman studies required much larger sampling area in comparison to the mc-AFM and therefore it can be expected that for the samples used in the mc-AFM measurements the photoconversion of E-1 to the metastable Z-1 is much higher. Half-life of the metastable Z-1 in the solid was calculated by following the gradual decrease in the intensity of the band centered at 1562 cm -1 and was found (ca. t 1/2 = 80 min.) to be similar to the half-life of the metastable E-1 in the solid state ( Figure S33a,b). Even though identification of the product of this thermal process was difficult due to the decomposition and low conversion to the metastable Z-1, appearance of some additional characteristic bands in the Raman spectrum ( Figure S32b,c) suggested formation of the stable Z-1 and thus competition of the rotary cycle.
In conclusion, in a stark contrast to the solution, molecular motor 1 in the solid operates as a fourstate chiroptical switch at ambient temperatures. Furthermore, in the solid, all four diastereoisomers are readily accessible in a sequential manner upon UV-light and heat treatment and thermal stabilities of both metastable E-1 and Z-1 are similar. Finally, it should be noted that even upon prolonged (>40 min.) irradiation of the identically prepared samples on quartz substrates no signs of photoisomerization or photodecomposition were observed, indicating crucial impact of the Au substrates on the photoisomerization process, most probably by facilitating melting of the samples upon local heating of the substrate.

CISS measurements in the solid state
Spin-dependent I-V curves were collected from each state of E and Z molecular motors by mc-AFM in ambient atmosphere. To determine the influence of the CISS effect on the electronic energy levels, the density of states as a function of bias voltage is required. This function is equivalent to dI/dV vs. V curves derived by analytical differentiation of the experimental I-V curves. The gap between the turning points of the resultant U-shaped curves gives the tunneling barrier. The energy gap was determined for the different configurations and field directions. A lower effective barrier means favorable spin selectivity represented by higher spin currents.