Controlling Photoconductivity in PBI Films by Supramolecular Assembly

Abstract Perylene bisimides (PBIs) self‐assemble in solution. The solubility of the PBIs is commonly changed through the choice of substituents at the imide positions. It is generally assumed this substitution does not affect the electronic properties of the PBI, and that the properties of the self‐assembled aggregate are essentially that of the isolated molecule. However, substituents do affect the self‐assembly, resulting in potentially different packing in the formed aggregates. Here, we show that the photoconductivity of films formed from a library of substituted PBIs varies strongly with the substituent and demonstrate that this is due to the different ways in which they pack. Our results open the possibility for tuning the optoelectronic properties of self‐assembled PBIs by controlling the aggregate structure through careful choice of substituent, as demonstrated by us here optimising the photoconductivity of PBI films in this way.


Photoconductivity and Wavelength Response Measurements
All samples for photoconductivity measurements were prepared on a glass microscope slide using a 3 x 3 mm mask. This was done placing a piece of Scotch tape on the glass slide and cutting out a 3 x 3 mm square using a scalpel. The sample could then be placed in the mask and dried. Once dried the mask was removed to leave a 3 x 3 mm square of the sample. Electrodag (Agar Scientific) silver paste was then put on either side of the sample making sure that the sample was touching the sides of the film but not covering it to maintain the 3 mm gap. Copper wires were then placed into the silver paste and dried. More silver paste was then applied and again left to dry. This was done to ensure good contact with the silver. Next epoxy resin was applied over the silver paste and a little over the wire and the glass. This prevents oxidation of the silver in the air, but also makes the silver/copper wire contacts more robust and less likely to break during measurements. For the solution 10 μL of sample was placed in the mask and dried in air.
All samples for photoconductivity were prepared and measured in triplicate to ensure reproducibility and the reported values in the main text are representative of the samples. Films were imaged by cross-polarised optical microscopy to check they had not cracked or crystallised upon drying.
Conductivity and wavelength response measurements were performed as we have previously reported. [1i] Directional conductivity measurements were performed as we have previously reported. [2]

Rheological Measurements
All rheological measurements and alignment experiments were performed using an Anton Paar Physica 301 or 101 rheometer. A cone and plate geometry was used for viscosity measurements and for the shear alignment measurements. All measurements were measured in triplicate to ensure reproducibility of reported results. Measurements were carried out at 25 °C.
Viscosity measurements: Measurements were carried out using a 50 mm cone from 0.1 s -1 to 1000 s -1 . Around 1 mL of solution was placed under the cone geometry.
Alignment experiment: Alignment was carried out using a 25 mm cone geometry. A piece of glass was secured to the bottom plate and 100 μL of solution was then placed on the glass and the cone lowered on the top. A shear rate of 10 s -1 was then used to shear align the sample. This was left overnight for the sample to dry under shear. This method was reported in detail in our previous work. [2]

UV-Vis Absorption Spectroscopy
All UV-vis absorption spectra were collected on a Cary 60 UV-vis spectrometer from Agilent Technologies. Wet UV-vis absorption spectra were collected using a Hellma 0.1 mm demountable quartz cuvette. The gels could be formed directly in the cuvettes and then measured. Dried film samples were made by pipetting 50 µL of gelator solution onto a glass microscope slide. This was then spread over the glass to make a thinner layer of sample and allowed to dry in air. To see the formation of the radical anion, or dianion, the thin film samples were irradiated with 365 nm LED for around 10 minutes and the UV-vis absorption spectra was recorded.

Optical Microscopy and Photos
Cross-polarised light optical microscope images were taken using a Nikon LV100 Eclipse Microscope fitted with an Infinity 2 colour camera. All other photographs were taken using an iPhone 7. No post-modification or processing was made to the images after being collected.

Small Angle Neutron Scattering (SANS)
SANS measurements were performed using the D11 instrument (Institut Laue -Langevin, Grenoble, France). A neutron beam, with a fixed wavelength of 6 Å and divergence of Δλ/λ = 9%, allowed measurements over a large range in Q [Q = 4πsin(θ/2)/λ] of 0.001 to 0.3 Å -1 , by using three sample-detector distances of 1.5 m, 8m, and 39 m. The solutions were prepared as described above, but replacing the H2O and NaOH with D2O and NaOD respectively. All measurements were carried out in 2 mm path length Hellma UV spectrophotometer grade, quartz cuvettes; the gels were prepared directly in these cuvettes. The cuvettes were housed in a temperature controlled sample rack during the measurements.
The data were reduced to 1D scattering curves of intensity vs. Q using the facility provided software LAMP. The electronic background was subtracted, the full detector images for all data were normalized and scattering from the empty cell was subtracted. The scattering from D2O was also measured and subtracted from the data. The data were normalized to absolute units using a 1mm thick water sample as secondary calibration standard, with a differential scattering cross section of 0.983 1/cm for the experimental settings used. Last, data were radially averaged to produce the 1D curves for each detector position. The instrument-independent data were then fitted to the models discussed in the text using the SasView software package version 3.1.2. [3] Cyclic Voltammetry (CV) CVs were collected using a three-electrode system and a Dropsens potentiostat with a glassy carbon working electrode, a Pt wire counter electrode and an Ag/AgCl reference electrode. The supporting electrolyte was 0.1 M NaCl in water and 0.1 M tetrabutylammonium hexafluoroborate (TBAHFB) in DMF. The measurements were scanned from 1.0 V to -1.0 V, firstly 5 times at a scan rate of 0.5 V/s and a step value of 0.01 V, then 5 scans were collected at 0.05 V/s with a step of 0.005 V. The broadness and resolution of the CVs sometimes changed between scans (assumed to be due to the high viscosity of solutions) although the absolute response and potential of the peaks did not change. Therefore, the clearest scan from the 5 scans at 0.05 V/s was selected for analysis.
For calibration, the E1/2 for the ferrocene-ferrocecium redox couple (Fc/Fc + ) in DMF was measured at 0.479 V vs. Ag/AgCl. Although not used for calibration, the E1/2 for the ferrocene carboxylic acid redox couple (Fc-COOH/Fc + -COOH) in water was measured at 0.616 V vs. Ag/AgCl.

Experimental ionisation potential (HOMO) and electron affinity (LUMO) calculations
For the determination of the electron affinity (EA)/ Lowest Unoccupied Molecular Orbital (LUMO) energy level vs. vacuum, the reduction maximum for the first reduction potential vs. Ag/AgCl was obtained from the CV data. For calculating the EA/LUMO energy in eV, potentials were then converted to vs. Fc/Fc + using an experimental value (+0.479 V), then to vs. SCE (+0.47 V), then to vs. SHE (-0.24 V), then to vs. vacuum (-4.28 V or -4.44 V) by using known literature conversions. [4] Therefore, the final calculation was: -EA = ELUMO (eV) = (Redmax + 0.479 V + 0.470 V -2.4 V -(4.28 V or 4.44 V)) For the aqueous solutions measured in the experiment, the solvent window for water (water oxidation) was reached at potentials > 0.9 V vs Ag/AgCl. Therefore, the oxidation of PBI expected to be around 1.5 V vs Ag/AgCl was inaccessible in aqueous solutions.
As the PBI materials in this study are used in photovoltaics, it was also deemed suitable to use the optical band gap to approximate the HOMO energy levels.
The ionisation potential (IP)/ Highest Occupied Molecular Orbital (HOMO) energy vs. vaccum was approximated by subtracting the optical band gap, Eg, from the EA/LUMO energy. Eg was calculated by using the onset of absorption from the UV-Vis spectrum (lonset) and converted to eV using the equation: Eg (eV) = 1242/l The approximate IP/HOMO energy was then calculated using the equation -IP = EHOMO (eV) = ELUMO -Eg.

S6
The HOMO and LUMO energies from solutions of PBI-X in both high pH water and dimethylformamide (DMF) were measured in order to compare the differences between the aggregated and dissolved material.

Theoretical calculations
The adiabatic ionisation potential (IP: PBI-XH + e --> PBI-XHor PBI-X -+ e --> PBI-X 2-) and electron affinity (EA: PBI-XH -+ e --> PBI-XH 2or PBI-X 2-+ e --> PBI-X 3-) of the singly-and doubly-deprotonated versions of the different PBI-X molecules (PBI-XH -/PBI-X 2-), representative of the (molecular) species present at high pH, were calculated in a ΔDFT fashion using the B3LYP [5] density functional and the COSMO [6] implicit solvation model, with a dielectric permittivity value of 80. The predicted IP and EA values were subsequently converted from the vacuum scale to that of the standard hydrogen electrode (SHE) by subtracting the experimentally reported absolute value of the SHE potential, using two different values from the literature 4.44 [4b] and 4.28 V [4a] respectively, and finally from the SHE scale to that of the Ag/AgCl electrode (saturated NaCl) by subtracting a further 0.197 V [7] . Ion-pair energies, the energy required to produce a pair of charged states sufficiently far separated that their electrostatic energy is negligible, also referred to as the fundamental or transport gap, are calculated from the differences of IP and EA values.
Two different basis-sets are employed; the double-ζ DZP [8] and the triple-ζ def2-TZVP [9] basissets, and all COSMO calculations include the outlying charge correction, which is found to change potentials by typically less than 0.1 V. All potential and ion-pair energy calculations, finally, are performed using Turbomole [10] 7.01.

SEM Imaging
High resolution imaging of the PBI hydrogel morphology was obtained using a Hitachi S-4800 FE SEM. PBI hydrogels were deposited onto conductive glass slides and attached to 15 mm Hitachi M4 aluminium stubs using an adhesive high-purity carbon tab before and air dried. The PBI samples were gold coated with gold for 1 minute at 25 mA a Emitech K550X automated sputter coater. Imaging was carried out at a working distance of 9 mm with an applied working voltage of 3 kV using a mix of upper and lower secondary electron detectors. The FE-SEM measurement scale bar was calibrated using certified SIRA calibration standards.

EPR Spectroscopy
All EPR data were recorded at X-band frequency (9.67 GHz) on a Bruker ELEXSYS E500 spectrometer equipped with an ER 4102ST-O optical transmission resonator. All measurements were collected on 0.5 mg/mL aqueous solutions in 1 mm o.d. quartz tubes. Film samples were deposited on 15 × 8 × 0.5 mm quartz plates and mounted in the middle of the resonator using a homemade sample holder. The holder was rotated 90° between irradiation and EPR data collection at various time intervals over 40 min. Field-swept spectra represent 5 scan averages collected over a 5 mT sweep width centered at 344.8 mT, with modulation frequency = 100 kHz, modulation amplitude = 0.05 mT, receiver gain = 60 dB, time constant = 40.96 s, sampling time = 10.24 s, and microwave power = 0.63 mW. Timescan measurements were performed at a fixed field position corresponding to the peak maximum in the first derivative spectrum under continuous irradiation with a 365 nm LED. Spin counts of solution samples were quantified by double integration of the first derivative spectrum and calibrated to a 0.5 mg mL -1 aqueous solution of TEMPO recorded under identical conditions. For solid state spectra (film) were curve fitted, and the fit double integrated to get the signal intensity for each time interval. This was converted to a spin count by comparison to the weak pitch standard measured under identical conditions. Solution samples for EPR were described as previously mentioned and transferred into soda glass capillary tubes with a diameter of 2 mm. Tubes were sealed at one end and samples filled 2 cm of the capillary tube.
Samples for solid EPR were prepared on pieces of 1 mm thick glass which was 8 mm by 20 mm. This could then be mounted inside the EPR spectrometer. The samples were then prepared the same as for photoconductivity measurements using a 3 x 3 mm mask. 10 µL of sample was pipetted into the mask and left to dry before the mask was removed.
Both solution and dried samples were irradiated with a 365 nm LED from LedEngin with an IsoTech CD laboratory power supply at 0.7 A and 3.7 V. Samples were irradiated until the signal intensity became stable.

Gelator
l max S0-S1 0-0 (nm) l onset (    Tables S7-S10 gives the IP, EA and ion-pair energies predicted for the different singly and doubly deprotonated PBIs as calculated with B3LYP/DZP and B3LYP/def2-TZVP. Calculations with both basis-sets give similar results and display a similar trend, more about which below, but use of the larger def2-TZVP basis-set, which for reasons of computational tractability has only been applied to a subset of molecules, results in the prediction of deeper IP values, slightly more shallow EA values and slightly larger ion-pair energies. Focussing on the def2-TZVP results, there is also a decent fit to the experimentally measured EA values under high pH conditions in table SX, as well as the experimental estimates of the IP values obtained by subtracting from the measured EA the adiabatic S1-S0 0-0 excitation energy, extracted from the UV-Vis absorption spectra. The calculations predict that the IP, EA and ion-pair energy values are very similar for the different singly-and doubly-deprotonated PBI-X molecules. A similar lack of change when varying the amino acid substituents is predicted for neutral PBI-X molecules in water (see Table S11). The differences in radical yield observed by EPR and UV-Vis between the different PBIs in solution is thus unlikely to be the result of differences in the fundamental electronic properties of the molecules. The main difference between the neutral and the singly-/doubly-deprotonated cases is the localization of the hole.
In line with what we observed in previous work for PBI-A [11] , the hole localizes, like an additional electron, on the PBI core in the neutral case and on the carboxylic acid group of the amino acid group after (partial) deprotonation. The only exception is PBI-Y, where both in the neutral and doubly deprotonated case, the hole localizes on the indole sidechain of the tryptophan amino acid substituent. The black data is before irradiation and the red date is after irradiation with a 365 nm LED. S17 Table S12. Percent of radical present in the PBI solution determined by EPR after irradiation with 365 nm.               The stability of the conductivity of the films was investigated by monitoring the evolution of the photocurrent while illuminating the films for a period before turning the lamp of again. The three samples that showed the presence of the dianion exhibited a longer grow in of the response, followed by a long decay of the current after the light was switched off. The current for the more responsive samples (PBI-A, PBI-H, PBI-S, and PBI-V, which have a similar UV-vis absorption spectrum to each other in solution) stabilized after illumination for 100 minutes. This current persisted longer than 10 hours after the light was switched off. However, with the less responsive samples, on irradiation, the current increased and then decreased again before stabilising after around 30 minutes. On turning off the light, the decay of the current then took a maximum of 100 minutes to decay to the dark current. PBI-F, PBI-L, PBI-W, and PBI-Y show this effect; these also have similar UV-vis absorption spectra (see main paper). This behavior could be repeated on the same film once the decay to the dark current had been reached, showing that the initial decrease under illumination is not due to film degradation. Figure S30. Comparison of the normalized X-band EPR spectra of the PBIs films at ambient temperature after achieving maximum signal intensity post irradiation with 365 nm (experimental conditions: frequency, 9.67 GHz; power, 0.63 mW; modulation, 0.05 mT).