Defying strain in the synthesis of an electroactive bilayer helicene† †Electronic supplementary information (ESI) available. CCDC 1864289 and 1864290. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c8sc04216k

Visible-light-induced oxidative cyclization of a phenanthrene framework overcomes immense strain to yield a bilayer perylene-diimide helicene.


Figure S2.
To show that the large extent of aryl surface overlap in PPDH inhibits racemization, we heated (-)-PPDH (1.7 mg) in diphenyl ether (0.75 mL) at 250 °C for 1 h. These are the HPLC traces of the solution before (a) and after (b) 1 h at 250 °C. This solution was injected in 10 µL aliquots onto a CHIRALPAK ® IB-3 column (4.6 mm I.D. × 250 mm, 3 µm), with 20% dichloromethane/hexanes flowing at 1 mL/min at room temperature. The small peak at ~6 min corresponds to the complete elution of diphenyl ether. There is no trace of (+)-PPDH, confirming that PPDH does not racemize under these conditions. Figure S3. From SCXRD, racemic PPDH stacks into heterochiral columns (red, M-PPDH; blue, P-PPDH). Solvent, the CH(C 5 H 11 ) 2 chains, and hydrogen atoms have been hidden to provide a clear view of the aryl surface. Figure S4. From SCXRD, the intermolecular junction between two molecules of PPDH consists of 24 pairs of overlapping π-bonded carbon atoms (shown in pink). (a) The four closest pairs, which approach to within twice the van der Waals radius of the carbon atom (i.e., 3.4 Å), are designated with black arrows. (b) Top view of the same PPDH molecules as in (a), only the uppermost and bottommost PDI subunits have been removed for clarity. The distances between the overlapping atoms (in Å) are indicated to the right of each pair, and the four nearest neighbors are underlined in bold. Free solvent, the CH(C 5 H 11 ) 2 chains, and hydrogen atoms have been hidden to provide a clear view of the aryl surface. Thermal ellipsoids are set at 30% probability. Figure S5. We define the bend angle of each PDI subunit in PPDH as the dihedral of the leastsquares-fit planes defined by the pink and blue naphthalene fragments. From SCXRD, the bend angle of one PDI subunit in PPDH measures 11°, whereas the bend angle of the other PDI subunit (planes not shown) measures 9°. Free solvent, the CH(C 5 H 11 ) 2 chains, and hydrogen atoms have been hidden to provide a clear view of the aryl surface. Thermal ellipsoids are set at 30% probability.

II. General Experimental Details
Crystallographic data for PPDH and PPDH-OPe are given in Section VIII. Slow vapor diffusion of acetonitrile into a solution of PPDH in anisole afforded bright red prisms. Slow vapor diffusion of acetonitrile into a solution of PPDH-OPe in a,a,a-trifluorotoluene afforded small orange rods. The crystals were mounted on MiTeGen Kapton loops (polyimide) using paratone oil. Data were collected at 100 K.
Quantum Mechanical Calculations: All quantum chemical calculations were performed using Jaguar, version 8.3, Schrodinger, Inc., New York, NY, 2014. 9 The geometries were optimized in the gas phase using the B3LYP functional and the 6-31G** basis set. For the optimized geometries of PPDH and PPDH-OPe, the associated absorption spectra were calculated using the TD-DFT method that is included in the Jaguar package. The B3LYP functional and the 6-31G** basis set were used in these calculations. All alkyl chains (CH[C 5 H 11 ] 2 and C 5 H 11 ) were modeled as methyl groups.
The effects of dispersion were also assessed in the geometric optimizations of PPDH and PPDH-OPe. In their energy benchmark study of 47 density functionals, Goerigk and Grimme emphasized the efficacy of DFT-D3 in modeling noncovalent interactions. 10 They regard Zhao and Truhlar's PW6B95, coupled with the DFT-D3 correction, as "the most robust and very accurate general purpose hybrid-functional." Therefore, the PW6B95-D3/6-31G** level of theory was also used to optimize the gas-phase geometries of PPDH and PPDH-OPe. These optimized geometries diverge substantially from the SCXRD structures of PPDH and PPDH-OPe, which closely resemble the geometries predicted by B3LYP/6-31G** ( Figure S23). Steric repulsion between the PDI subunits evidently predominates in these systems, which makes the B3LYP functional an appropriate choice for the calculation of strain (see Section VII).

III. Synthesis and Characterization
Scheme S1. Synthesis of PPDH and 5PPD Scheme S2. Synthesis of PPDH-OPe 3,6-Dibromophenanthrene (S1): trans-4,4'-Dibromostilbene (0.362 g, 1.07 mmol, 1 eq), iodine (0.603 g, 2.38 mmol, 2.22 eq), and propylene oxide (2.0 ml, 29 mmol, 27 eq) were combined with 310 mL of benzene in a 320-mL quartz round-bottom flask and sparged with nitrogen for 10 minutes. The flask was placed in a Rayonet photoreactor (The Southern New England Ultraviolet Company) with sixteen 300 nm lamps and stirred under UV light for 8 h. This reaction mixture was combined with another batch that started with 0.314 g of dibromostilbene. The solvent was removed under reduced pressure. The solid was purified by hot recrystallization from hexanes to yield 0.499 g of white needles (1.48 mmol, 74% over two batches). All spectra matched those reported in the literature. 2 S2: S1 (0.107 g, 0.318 mmol, 1 eq), bis(pinacolato)diboron (0.176 g, 0.692 mmol, 2.18 eq), potassium acetate that had been dried in a 200 °C oven (0.243 g, 2.48 mmol, 7.79 eq), and [1,1'bis(diphenylphosphino)ferrocene]dichloropalladium (0.0163 g, 0.0223 mmol, 7.00 mol%) were placed in an oven-dried 10-mL Schlenk flask, then evacuated and back-filled with nitrogen three times. In a separate oven-dried 10-mL round-bottom flask, 2 mL of anhydrous 1,4-dioxane were sparged for 8 min, then transferred to the reaction mixture and sparged for 3 min. The Schlenk flask was sealed with a glass stopcock and heated to 80 °C overnight, at which point it was added to 50 mL of deionized water. The aqueous layer was extracted with 3 x 50 mL of ethyl acetate. The organic layer was dried with MgSO 4 , filtered, and the solvent removed with a rotary evaporator. Purification by column chromatography (SiO 2 , gradient from 100% hexanes to 100% dichloromethane) afforded the white solid S2 (0.0406 g, 0.0944 mmol, 30%). 1

S3
: PDIBr (0.152 g, 0.196 mmol, 2.12 eq), S2 (0.0398 g, 0.0925 mmol, 1 eq), K 2 CO 3 (0.300 g, 2.17 mmol, 24.4 eq), and [1,1'-bis(diphenylphosphino)ferrocene]dichloropalladium (0.0088 g, 0.012 mmol, 13 mol%) were placed in a 10-mL Schlenk flask, then evacuated and back-filled with nitrogen three times. In a separate 10-mL round-bottom flask, 3 mL of tetrahydrofuran and 1 mL of deionized H 2 O were sparged with nitrogen for 10 min, then transferred to the reaction mixture and sparged for 4 min. The Schlenk flask was sealed with a glass stopcock and heated at 75 °C overnight, at which point it was added to 35 mL of deionized water. It was extracted with dichloromethane until the aqueous layer turned colorless. The organic layer was dried with Na 2 SO 4 , filtered, and the solvent removed with a rotary evaporator. Purification by column chromatography (SiO 2 , gradient from 100% hexanes to 100% dichloromethane) afforded the dark red solid S3 (0.133 g, 0.0847 mmol, 92%). 1  PPDH and 5PPD: PPD (0.0308 g, 0.0196 mmol, 1 eq), iodine (0.0322 g, 0.127 mmol, 6.48 eq), and K 2 CO 3 (0.545 g, 3.94 mmol, 201 eq) were dissolved in 98 mL of benzene in a 150-mL round-bottom flask. The solution was sparged with nitrogen for 30 min and left under positive pressure of nitrogen while being irradiated by two 55 W CFLs for 24 h at 30 °C (the temperature to which the light bulbs heated the solution) in a pristine oil bath. The solvent was removed by rotary evaporation and the material was loaded onto a small silica plug and dried with air. The plug was flushed with acetonitrile (40 mL) to remove iodine and benzene (but not the products, which are insoluble in acetonitrile and stay on the baseline). The mixture of products was brought down with 9:1 (v/v) dichloromethane/ethyl acetate and the solvent was removed by rotary evaporation. 1 H NMR was taken of this mixture in C 2 D 2 Cl 4 at 393 K ( Figure S7). To isolate the products, the mixture was loaded onto a plug again and flushed with 9:1 (v/v) dichloromethane/hexanes, then dichloromethane. These dichloromethane washes contained only 5PPD (0.0108 g, 0.0689 mmol, 35% isolated yield). PPDH was brought down with 9:1 (v/v) dichloromethane/ethyl acetate and the solvent removed by rotary evaporation to give a red solid (0.0193 g, 0.0123 mmol, 63% isolated yield). PPDH: 1  This reaction was repeated with PPD (0.0331 g, 0.0211 mmol, 1 eq), iodine (0.0371 g, 0.146 mmol, 6.94 eq), and K 2 CO 3 (0.634 g, 4.59 mmol, 218 eq) in 106 mL of benzene at 70 °C for 24 h in a pristine oil bath. Isolation and purification followed the same procedure as above to give 0.0274 g (0.0175 mmol, 83%) of PPDH. This reaction was repeated with PPD (0.0268 g, 0.0171 mmol, 1 eq), iodine (0.0291 g, 0.115 mmol, 6.72 eq), and K 2 CO 3 (0.476 g, 3.44 mmol, 202 eq) in 90 mL of chlorobenzene at 110 °C for 24 h in a pristine oil bath. Isolation and purification followed the same procedure as above to give 0.0242 g (0.0154 mmol, 91%) of PPDH. Figure S7. 1 H-NMR spectra (500 MHz, C 2 D 2 Cl 4 , 393 K) of the product mixtures (after an acetonitrile plug to remove iodine and benzene) resulting from the oxidative photocyclization of PPD at different temperatures. The integral shown for PPDH corresponds to two protons and the integral shown for 5PPD corresponds to one proton.
3,6-Dibromophenanthrene-9,10-quinone (S3): This molecule was synthesized according to a published procedure. 3 All spectra matched those previously reported. S4: 3,6-Dibromophenanthrene-9,10-quinone S3 (5.45 g, 14.9 mmol, 1 eq), Na 2 S 2 O 4 (25.9 g, 149 mmol, 10.0 eq), and tetrabutylammonium bromide (4.82 g, 15.0 mmol, 1.00 eq) were placed in a 500-mL round-bottom flask. Tetrahydrofuran (110 mL) and deionized water (110 mL) were added. The flask was capped and shaken for 6 min. 1-Bromopentane (8.1 mL, 65 mmol, 4.4 eq) was added, followed by KOH (22.0 g, 392 mmol, 26.3 eq) in 110 mL of water. The mixture became dark. It was allowed to stir for 48 h, after which it was judged complete by TLC (95:5 [v/v] hexanes/ethyl acetate). The aqueous layer was extracted with 3 x 200 mL of ethyl acetate. The organic layers were combined and washed with water (2 x 200 mL), brine (1 x 100 mL), dried with Na 2 SO 4 , decanted, and the solvent removed with a rotary evaporator to yield a brown oil. Ethanol was added to precipitate the product, which was further washed with ethanol, leaving 4.59 g (9.03 mmol, 61%) of a pale yellow solid. This product contained trace (<5% by NMR) S4-deO, a molecule that lacks one oxygen. S4-deO could not be removed from S4 by silica gel column chromatography or recrystallization, so the product was carried forward. The monodeoxygenated impurity was removed by preparative HPLC after the oxidative photocyclization of PPD-OPe. An analytically pure sample of S4 and (and S4-deO, whose 1 H-NMR spectrum is included in Section VI) were obtained by preparative TLC (cyclohexane). 1

S5:
A 50-ml Schlenk flask was charged with S4 (0.982 g, 1.93 mmol, 1 eq), tetrahydrofuran (36 mL), and N,N,N′,N′-tetramethylethylenediamine (0.64 mL, 4.3 mmol, 2.2 eq). The flask was immersed in an acetone/dry ice bath. After being cooled for 15 min, 1.59 M n-butyllithium in hexanes (2.7 mL, 4.3 mmol, 2.2 eq) was added dropwise over 9 min and the solution was allowed to stir for an hour. 2-Isopropoxy-4,4,5,5-tetramethyl-1,3,2-dioxaborolane (0.91 mL, 4.4 mmol, 2.3 eq) was added and the solution was stirred cold for 20 min, then warmed up to room temperature. The reaction was monitored by TLC (95:5 [v/v] hexanes:ethyl acetate) and judged complete after 1 h. The reaction mixture was poured into saturated aqueous NH 4 Cl (100 mL) and extracted with ethyl acetate (3 x 100 mL). The organic layer was dried with MgSO 4 , filtered, and the solvent removed with a rotary evaporator to yield a slowly solidifying brown solid. The solid was recrystallized twice from ethanol to give 0.62 g (1.04 mmol, 54%) of white crystals. 1  PPD-OPe: PDIBr (0.498 g, 0.640 mmol, 2.17 eq), S5 (0.178 g, 0.295 mmol, 1 eq), K 2 CO 3 (1.32 g, 9.53 mmol, 32.3 eq) and [1,1'-bis(diphenylphosphino)ferrocene]dichloropalladium (48.1 mg, 0.0657 mmol, 22.3 mol%) were placed in a two-neck 50-mL round-bottom flask fitted with a reflux condenser. The flask was evacuated and backfilled with nitrogen three times. In a separate flask, tetrahydrofuran (18 mL) and H 2 O (2 mL) were sparged for 30 min with nitrogen. The solvents were transferred into the flask with the solids by syringe and sparged for 15 min. The reaction mixture was heated to reflux for 16 h, at the end of which it was added to 50 mL of deionized water. The aqueous layer was extracted with dichloromethane until it became clear (~100 mL). The organic layer was dried with MgSO 4 , filtered, and the solvent removed with a rotary evaporator. Purification by column chromatography (SiO 2 , gradient from 100% hexanes to 90% dichloromethane) afforded a red solid (0.348 g, 0.200 mmol, 68%). 1  PPDH-OPe: PPD-OPe (0.0291 g, 0.0167 mmol, 1 eq), iodine (0.0285 g, 0.112 mmol, 6.73 eq), and K 2 CO 3 (0.460 g, 3.33 mmol, 200 eq) were dissolved in 78 mL of benzene in a 150-mL round-bottom flask. The solution was sparged with nitrogen for 30 min and left under positive pressure of nitrogen while being irradiated by two 55 W CFL light bulbs for 76 h in a pristine oil bath. The amount of mono-cyclized intermediate decreased over this time, as observed by TLC (4:1 [v/v] dichloromethane/hexanes). However, the amount of decomposition also increased, so the reaction was halted. The solvent was removed by rotary evaporation and the material was loaded onto a small silica plug and dried with air. The plug was flushed with acetonitrile (40 mL), followed by 9:1 (v/v) dichloromethane/hexanes, then dichloromethane. These dichloromethane washes contained the mono-cyclized intermediate and decomposition (which have a combined mass of 4 mg. The decomposition product has broad, unidentifiable peaks in its 1 H-NMR spectrum and a mass that corresponds to loss of the pentyl groups). PPDH-OPe was brought down with 9:1 (v/v) dichloromethane/ethyl acetate and the solvent removed by rotary evaporation to give a red solid (0.0256 g, 88%). Due to the difficulty of separating S4 from S4-deO on a large scale, PPDH-OPe contains a small amount of the [7]helicene PPDH-OPe-deO (<5% of the product by NMR [ Figure S8]). The separation between such similar molecules by HPLC is poor and the recovery is 76%, with the rest remaining in mixed fractions ( Figure S9). 1 H NMR (500 MHz, C 2 D 2 Cl 4 , 393 K) δ 10.26 (s, 2H), 9.47 (d, J = 9.0 Hz, 2H), 9.14 (d, J = 8.9 Hz, 2H), 9. Figure S11. UV-visible absorbance spectra of PPDH, PPDH-OPe, and NPDH in cyclohexane (10 µM, 1 cm path length) and fluorescence spectra of PPDH and PPDH-OPe in cyclohexane (3 µM, λ ex = 410 nm). Inset shows PPDH (left vial) and PPDH-OPe (right vial) in cyclohexane under a UV lamp emitting ~254 and ~365 nm light.

VII. DFT-Optimized Molecular Structures and TD-DFT Excited State Calculations Part A: Calculating Strain Energy
We minimized the geometries of PPDH, 5PPD, and PPPD by DFT at the B3LYP/6-31G** level of theory. We calculated the strain energy of the helicenes by the formula: The total energies (in hartrees) of the six phenanthrene-bridged PDI-dimers are provided on pages 34, 43, 46, 50, 58, and 61. As an isomer of PPDH and 5PPD, PPPD is a good reference because it is nearly planar and, therefore, virtually unstrained. The CH(C 5 H 11 ) 2 and C 5 H 11 chains were modeled as methyl groups to simplify the calculations.      Figure S18. Highest-and lowest-unoccupied molecular orbitals of PPDH-OPe by DFT (B3LYP/6-31G**). Orbital isosurfaces are illustrated at 0.05 electrons Bohr -3 . We substitute methyl groups for the CH(C 5 H 11 ) 2 and C 5 H 11 chains to simplify the calculations.     Figure S23. Optimization in the gas phase at the B3LYP/6-31G** level of theory returns geometries of PPDH and PPDH-OPe (depicted in red) that resemble the corresponding structures from SCXRD (orange). In contrast, optimization at the PW6B95-D3/6-31G** level of theory gives highly compressed bilayers (blue). We substituted methyl groups for the CH(C 5 H 11 ) 2 and C 5 H 11 chains to simplify the DFT calculations. These alkyl groups have been hidden in the structures above to provide an unobstructed view of the aryl surfaces.

VIII. Single-crystal X-ray Diffraction Data
Crystallographic data corresponding to PPDH have been deposited with the Cambridge Crystallographic Data Centre (CCDC #1864290).