Synthesis of Nonplanar Graphene Nanoribbon with Fjord Edges

As a new family of semiconductors, graphene nanoribbons (GNRs), nanometer-wide strips of graphene, have appeared as promising candidates for next-generation nanoelectronics. Out-of-plane deformation of π-frames in GNRs brings further opportunities for optical and electronic property tuning. Here we demonstrate a novel fjord-edged GNR (FGNR) with a nonplanar geometry obtained by regioselective cyclodehydrogenation. Triphenanthro-fused teropyrene 1 and pentaphenanthro-fused quateropyrene 2 were synthesized as model compounds, and single-crystal X-ray analysis revealed their helically twisted conformations arising from the [5]helicene substructures. The structures and photophysical properties of FGNR were investigated by mass spectrometry and UV–vis, FT-IR, terahertz, and Raman spectroscopic analyses combined with theoretical calculations.


Synthesis of P1
To a Schlenk tube was added compound 10 (300 mg, 0.548 mmol), bis(tri-tertbutylphosphine)palladium(0) (7.00 mg, 13.7 µmol), 1.4 mL of degassed aqueous tripotassium phosphate solution (3 M), and 1.4 mL of degassed THF under argon. The mixture was stirred at 50 ℃ for 24 h. Then, bromobenzene (86.0 mg, 0.548 mmol) was added, and the resulting mixture was stirred for 12 h. Subsequently, phenyboronic acid (66.8 mg, 0.548 mmol) in 0.5 mL of degassed THF was added, and the resulting mixture was stirred for another 12 h. Then, 20 mL of water was added before the aqueous phase was extracted by dichloromethane (three times). The combined organic phases were washed with water and dried over magnesium sulfate. After the removal of the solvent under reduced pressure, the residue was dissolved in 5 mL of dichloromethane followed by precipitation into 100 mL of methanol. The white precipitates were collected via filtration and dried under vacuum. The white solid was further washed by Soxhlet extraction with boiling acetone for 2 days, and then dried under reduced pressure at room temperature to afford P1 as a white solid (184 mg, 98%). FT-IR (powder) 3085, 3052, 3033, 2958, 2902, 2866, 1514, 1477, 1461, 1440, 1391, 1361, 1268, 1199, 1109, 1028, 1006, 581 cm -1 .

Synthesis of FGNR
A solution of P1 (20.0 mg) in 25 mL of unstabilized dichloromethane was degassed by argon bubbling for 20 min. To the degassed solution was added dropwise a degassed suspension of iron(III) chloride (324 mg) in 3 mL of nitromethane. After stirring at 45 °C for 48 h, the reaction was quenched by addition of 10 mL of methanol. The precipitates were collected via filtration. Then, the collected solid was re-dispersed in 10 mL of THF followed by adding dropwise into 100 mL of methanol. The precipitates were collected via filtration followed by washing intensively with methanol and dried under vacuum to afford FGNR as a dark red solid (17.4 mg, 88% yield). FT-IR (powder) 2958, 2924, 2866, 1702, 1598, 1461, 1397, 1365, 1341, 1249, 1200, 1123, 1073, 587 cm -1 .

Isomerization processes of 1 and 2
To evaluate the isomerization processes of 1 and 2, DFT calculations were performed. As illustrated in Figure S3, the interconversion between [M]-and [P]-1 was proposed to adopt a transition state where terminal benzene rings of the [5]helicene substructure are oriented face-to-face because of the bulky tert-butyl groups ( Figure S3a). Accordingly, the isomerization barrier was estimated to be 36.0 kcal/mol which is by 13 kcal/mol higher than that of [5]helicene. 2       The solvent-free sample preparation allows FGNR to be detected as radical cation under the MALDI condition, similar to polycyclic aromatic hydrocarbons (PAHs) and fullerenes. [3][4][5] The observed mass result of FGNR was used directly to evaluate cyclodehydrogenation efficiency by comparing the number of hydrogen atoms lost during the Scholl reaction, derived from mass difference between corresponding peaks of P1 and FGNR in their MALDI-TOF MS, with the theoretical number of hydrogen atoms removed. 6   Table S1. Estimation of the cyclodehydrogenation efficiency based on MALDI-TOF MS. FT-IR spectra of P1 and FGNR. Blue dashed rectangle indicates a peak from SOLO mode at 870 cm -1 (wagging of an isolated aromatic C-H bond neighbored by two C-C bonds, orange-colored in inset of Figure S9b), which is absent in the spectrum of P1.

Electrochemical properties of 1 and 2
Figure S10. Cyclic voltammograms of (a) 1 and 2; (b) ferrocene. Only the oxidation process was observed for both 1 and 2. HOMO was estimated with ferrocene as an external standard. The oxidation potential of ferrocene was regarded as −4.8 eV from the vacuum level.

X-ray crystallographic analysis of 1 and 2
The X-ray crystallographic coordinates for structures reported in this work have been deposited at the Cambridge Crystallographic Data Centre (CCDC), under deposition number 2058017 (1)    Helicity of model compound 1 is obvious as only one [5]helicene subunit embedded. However, the situation becomes more complex for model compound 2 with two more incorporated [5]helicene subunits. Both NMR and chiral HPLC spectrum of 2 indicated the existence of multiple conformers, but only [M,M,M]-2 and [P,P,P]-2 were found in its crystal structures. In order to elucidate the conformations of compound 2, we performed the theoretical optimization (DFT-B3LYP/ 6-31G(d,p)

DFT calculations
DFT computations were carried out by using the Gaussian 16 software package. 7 Different short oligomers and infinite GNR (using periodic boundary condition) were computed at the DFT level of theory with the HSE06 functional and 6-31G(d) basis set. Geometry optimization are followed by frequency calculations to obtain the IR spectra and by TD-DFT single point calculations to obtain the absorption spectra. In order to create the repeating unit for the infinite FGNR, FGNR-monomer (as shown in Figure S14b) is considered. tert-Butyl substituent is considered for all the oligomers and FGNR.

Electronic properties
The extension of π-conjugation reduces the energy gap. The energy gap decreases from 2.65 to 2.08 eV going from the 1 to FGNR-tetramer. The band gap of FGNR is similar to the energy gap obtained for FGNR-tetramer, with a value of 1.93 eV, suggesting reaching of saturation already for short oligomers, as shown in Figure S15. The optical energy gaps of 1 and 2 deduced from the absorption onset are 2.52 and 2.25 eV, respectively. It should be noted that 1 displays a smaller gap compared to its analog with a flat-lying geometry (2.64 eV), 8 indicating the impact of distortion on electronic property. The shapes of the frontier orbitals of 1, 2, and FGNR-tetramer are reported in Figure S16. All the HOMO orbitals propagate along the length direction, while the LUMO are oriented more along the width direction. Figure S16. The shapes of the frontier orbitals of 1, 2, and FGNR-tetramer.

Optical properties
The absorption spectra of 1, 2, and FGNR-tetramer have been calculated and is reported in Figure S17. A strong red shift is observed going from 1 to FGNR-tetramer, with the lowest absorption peak shifting from 466 to 600 nm. In specific, a total shift of 72 nm going from 1 to 2 and of 62 nm from 2 to FGNRtetramer is demonstrated. The longest wavelength absorption is described by a HOMO to LUMO transition for all oligomers. The absorption spectra are quite complex and show many peaks. Table S3 summarises the simulated transitions of 1, 2, and FGNR-tetramer.

Terahertz spectroscopic study of FGNR
Ultrafast optical pump−terahertz probe (OPTP) spectroscopy was employed to measure the photoconductivity of FGNR. The working principle of OPTP spectroscopy and its applicability for unveiling the charge transporting properties in nanostructured materials have been widely reported previously. [9][10][11] Drude-Smith (DS) model and fitting The Drude-Smith (DS) model, modified from the classic Drude model, has been employed intensively to characterize the charge transport properties in nanostructured semiconductors (such as graphene nanoribbons 12,13 ). In the DS model, a parameter c is introduced to characterize the backscattering probability, e.g., due to structural confinement. The frequency-resolved conductivity spectra of FGNR fitted by the DS model using the expression: where τ, and are the charge scattering time, the plasma frequency, and vacuum permittivity, respectively.

Reduced mass of photogenerated charge carriers
Due to the presence of tert-butyl substituents at the peripheral sites which induces steric hindrance, the geometry of FGNR is distorted, with a tilt angle of the edge phenyl groups of about 40-45°, as shown in the unit cell ( Figure S18) that applied for the calculation of infinite FGNR. This, in turn, affects the transport properties of FGNR. From the band dispersion, which are reported in Figure S19, we computed the reduced mass around the VB maximum (VBM) and CB minimum (CBM), as well as the VBM and CBM orbitals. As result, we obtained a value of 0.95 for both mh* (effective mass of the electron) and me* (effective mass of the hole). Furthermore, we estimated the effective reduced mass of charge carriers m*, by considering the averaged values for both charges, with the equation: * * * . The intrinsic charge mobility µ ( * ) was estimated to be 104 ± 3 cm 2 V -1 s -1 . Note that, this mobility value is slightly lower than GNRs with more planar structure and similar charge scattering time (20~30 fs). 13,14 This is mainly due to a relatively large effective mass of the charge carriers in FGNR, originating from its non-planar geometry.