Structure–Property Relationships for the Electronic Applications of Bis-Adduct Isomers of Phenyl-C61 Butyric Acid Methyl Ester

Higher adducts of a fullerene, such as the bis-adduct of PCBM (bis-PCBM), can be used to achieve shallower molecular orbital energy levels than, for example, PCBM or C60. Substituting the bis-adduct for the parent fullerene is useful to increase the open-circuit voltage of organic solar cells or achieve better energy alignment as electron transport layers in, for example, perovskite solar cells. However, bis-PCBM is usually synthesized as a mixture of structural isomers, which can lead to both energetic and morphological disorder, negatively affecting device performance. Here, we present a comprehensive study on the molecular properties of 19 pure bis-isomers of PCBM using a variety of characterization methods, including ultraviolet photoelectron spectroscopy, thermal gravimetric analysis, differential scanning calorimetry, single crystal structure, and (time-dependent) density functional theory calculation. We find that the lowest unoccupied molecular orbital of such bis-isomers can be tuned to be up to 170 meV shallower than PCBM and up to 100 meV shallower than the mixture of unseparated isomers. The isolated bis-isomers also show an electron mobility in organic field-effect transistors of up to 4.5 × 10–2 cm2/(V s), which is an order of magnitude higher than that of the mixture of bis-isomers. These properties enable the fabrication of the highest performing bis-PCBM organic solar cell to date, with the best device showing a power conversion efficiency of 7.2%. Interestingly, we find that the crystallinity of bis-isomers correlates negatively with electron mobility and organic solar cell device performance, which we relate to their molecular symmetry, with a lower symmetry leading to more amorphous bis-isomers, less energetic disorder, and higher dimensional electron transport. This work demonstrates the potential of side chain engineering for optimizing the performance of fullerene-based organic electronic devices.


Naming and Structure Assignments of Bis-PCBM Isomers
The label number of each carbon atom on C60 cage is shown in Figure S1.As all double bonds on C60 cage are symmetrically equivalent, the first addend crosses carbon atoms 1 and 9 by default.Cis isomers: two side chains are located at the same hemisphere; e isomers: the second side chain in the equator; trans isomers: the second side chain is located at different hemisphere.

Figure S1
. All carbon atom labels on C60 cage for the naming system of bis-PCBM isomers.
Table S1.Isomer structure assignments to the HPLC fractions with the first addend at the 1,9 position. 1The point symmetry* of each isomer is also shown.Table S2.HOMO, LUMO and HOMO-LUMO gap energies of the bis-PCBM isomers from density functional theory calculation (T), isomer solution using cyclic voltammetry and UVvis (S) and isomer film using UPS and UV-vis (F).The theoretical HOMO energy is calculated using DFT and the LUMO energy was estimated by adding the HOMO energy to the HOMO-LUMO gap which is the transition energy of the first singlet excitation calculated using TD-DFT using the same functional and basis set.The experimental HOMO energies are estimated from the oxidation potential of fullerenes in solution measured by CV and of thin films measured using UPS.The experimental LUMO energies are estimated, for fullerenes in solution, from the reduction potential measured using CV and, for films, from the UPS measured HOMO energy plus the thin-film energy gap measured from the UV-vis spectra.

FTIR/NMR
The FTIR spectra of bis-PCBM isomers were measured in air atmosphere in transmission mode using a Bruker Alpha P FTIR spectrometer with DGTS detector.The spectra were recorded from 700 to 3800 cm -1 with 1 cm -1 step size. 1 H NMR spectra were recorded on a Bruker AVNEO_600 spectrometer at 298 K using tetramethylsilane (TMS) as an internal standard.
Deuterated Chloroform was used for the NMR measurements.

Single Crystal X-ray Diffraction
Single crystals of isomers 2.3 and 3.3.2were obtained using antisolvent evaporation method. 3,4 Aproximately 5 mg of sample was placed in 2 ml of IPA in a small vial and heated under constant stirring on a hotplate to 110 °C, which is between the boiling point of IPA and CB.
CB (ca. 1 ml) was added dropwise along the inner vial wall until all the sample had dissolved and the solution then cooled slowly to room temperature.The cooled solution was then filtered through a 0.45 µm membrane filter into a second vial with a sealed cover.The cover was pierced with a small hole to allow slow evaporation of the solvent.After several days, crystals were found on the inner wall.At this point most of the solvent mixture was removed using a teat pipette and the crystals washed with IPA.The crystals were then transferred onto a silicon slide for characterization.Crystals for isomer 7 were easy to obtain, and were precipitated from toluene solution directly without an antisolvent.
Single-crystal X-ray diffraction data were collected at the UK National Crystallography Service at the University of Southampton on a Rigaku, 007-HF 4-circle diffractometer fitted with a Rigaku, HyPix 6000 hybrid pixel detector.Data were collected at 100 K using Cu-Kα radiation (λ = 1.5418Å).The structures were solved with using direct methods in SHELXTL 5 and refined within SHElXL 6 within the WINGX Gui. 7 Molecular graphics and C60 centroid-C60 centroid distances were obtained using CrystalMaker. 8gure S6 shows the X-ray single crystal structure of isomers 2.3, 3.3.2and 7 with thermal ellipsoids representing 50% probability.

Voltage Loss Analysis
The voltage losses were investigated quantitatively using electroluminescnce (EL) measurements along with sub-band-gap external quantum efficiency (EQE) measurements.
[11] The normalized EL and EQE spectra for the P3HT:fullerene devices is shown in Figure S7.
The EQE are composed of directly measured quantum efficiency and the quantum efficiency determined from the EL spectra, from which the radiative recombination Voc,rad can be calculated based on the method described in detail by Yao et al. 9 Then the ΔVoc,abs and ΔVoc,nr values can be calculated by Voc,sq -Voc,rad and Voc,rad -Voc .
The EQE spectra were measured using a grating spectrometer (CS260-RG-4-MT-D) to create monochromatic light combined with a tungsten halogen light source.The monochromatic light was chopped at 300 Hz, and a Stanford Research System SR380 lock-in amplifier with an internal transimpedance amplifier of 10 6 V/A was used to detect the photocurrent.Long pass filters at 610, 715, 780, 850, and 1000 nm were used to filter out the scattered light from the monochromator.The spectra were taken from 300 to 1100 nm and calibrated by a silicon photodiode.
EL was measured using a Shamrock 303 spectrograph combined with an iDUS InGaAs array detector cooled to -90 °C.The obtained EL spectra intensity was calibrated with the spectrum from a calibrated halogen lamp and the raw spectrum was corrected by subtracting a dark spectrum and using a calibration file that corrects for the detector sensitivity at different wavelengths.A Keithley 2450 source meter was used to provide power to drive the samples.SEM images were recorded on an FEI Inspect-F scanning electron microscope.(B) The AFM images of the PBDB-T:fullerene blend films, exhibiting amorphous and smooth property. 12The blends with PCBM, bis-mix and crystalline isomer-5.2.2 show slightly higher roughness of 18 nm, 16 nm, and 14 nm.The blends with amorphous isomers 3.2.1,3.2.2 and 5.1 show a smaller roughness of 7.3 nm, 8.5 nm, and 11.1 nm.

Model System
For each of the crystal structures studied, the centroid positions of the C60 cages were first found within the unit cell.These are referred to as sites.3×3×3 supercells was constructed using these site positions.Pairwise distances,   , between sites were computed using a function from the python package MDAnalysis. 13,14 his function handles the periodic boundaries of the supercell by applying the minimum image convention.The electron transfer integrals between molecule pairs were estimated using Values for the parameters  and  were chosen following previous work by Steiner and colleagues and are given in Table S5. 15Because the charge transport simulation program assumes an external electric field oriented along the z-axis, using the unmodified site coordinates yields the mobility in the a×b direction of the crystal.To study transport in directions b×c and c×a, the site positions were rotated around the origin such that the b-c and c-a crystal planes were parallel with the x-y plane.

Charge Transport Simulation
Mobilities were found by solving the steady state Master Equation for the system: which can be expressed in matrix form as Where  → is the rate constant at which charges hop from site i to site j, and   is the occupation probability of site i and it is assumed that all   are close to 0. Hopping rates were computed using semi-classical Marcus theory: The term ∆E → describes the difference in energy between sites i and j, ∆E → =   −   where   is the energy of an electron in site  .In the absence of energetic disorder, this difference arises solely due to the presence of the external electric field, which modifies ∆E → like: where the negative sign indicates we are studying electron transport.The system of simultaneous equations described by Equation 3 was solved by singular value decomposition, using a function from the GNU Scientific Library. 16This yields occupation probabilities of the sites, .Using these, we can compute the drift velocity of the charges according to  = ∑      ⃗  • î ≠ , (6)   from which the mobility is found using Table S6.Electron transport mobilities computed using total reorganisation energy of 0.5 eV, i.e. external contribution of 0.3 eV.Unit for mobility is cm 2 V -1 s -1 .

Figure S2 .
Figure S2.The molecular structures optimized using DFT (B3LYP/6-311G(2df, 2pd)) and wavefunction density distribution of the bis-PCBM isomers with the first side chain at the north pole.Isosurfaces of the HOMO and LUMO wavefunction are also exhibited.

Figure S3 .
Figure S3.(a-f) The UV-vis spectra of the different types of bis-PCBM isomer in solution (toluene) and thin film. 2 (g) The comparison among the HOMO-LUMO gaps from the theoretical calculation and the UV-vis measurement of the solution and film isomers.

Figure S5 .
Figure S5.(a) FTIR spectra of the isomers before and after heating at 120 °C under N2.(b) The

Figure S7 .
Figure S7.(a) Normalized EL and EQE spectra for the P3HT:fullerene devices, (b) EL peak energy as a function of open circuit voltage.The EQE are composed of directly measured quantum efficiency and the quantum efficiency determined from the EL spectra.

Figure S8 .Figure S9 .
Figure S8.(a) Extracted gaussian width of the EL spectra and (b) peak EL energy for organic photovoltaic devices based on P3HT polymer blended with different bis-isomers.

Table S3 .
Extracted device parameters from current-voltage characteristics of

Table S5 .
Input parameters for charge transport simulations