Increasing Photovoltaic Performance of an Organic Cationic Chromophore by Anion Exchange

Abstract A symmetrical cyanine dye chromophore is modified with different counteranions to study the effect on crystal packing, polarizability, thermal stability, optical properties, light absorbing layer morphology, and organic photovoltaic (OPV) device parameters. Four sulfonate‐based anions and the bulky bistriflylimide anion are introduced to the 2‐[5‐(1,3‐dihydro‐1,3,3‐trimethyl‐2H‐indol‐2‐ylidene)‐1,3‐pentadien‐1‐yl]‐1,3,3‐trimethyl‐3H‐indolium chromophore using an Amberlyst A26 (OH− form) anion exchanger. Anionic charge distribution clearly correlates with device performance, whereby an average efficiency of 2% was reached in a standard bilayer organic solar. Evidence is given that the negative charge of the anion distributed over a large number of atoms is significantly more important than the size of the organic moieties of the sulfonate charge carrying group. This provides a clear strategy for future design of more efficient cyanine dyes for OPV applications.


General Information
General Notation for this Work Figure S1. Overview of the used notation of the synthesized compounds for this work.

Anion Exchange Procedure General Loading of the Resin
An appropriate amount of the wet anion exchange resin Amberlyst® A26 (OHform) was packed in a glass column (0.5 cm diameter). The column bed was then equilibrated with water until a constant pH value was reached. Then a 1% water solution of the corresponding acid was passed through the column until the eluate had reached the same pH as the original acid solution. During anion loading a change in colour of the resin from pink to pale yellow was obtained.
The pH changed from approximately 5 to 2. Then the resin was washed with water until constant pH was reached and equilibrated with the selected solvent media used for anion exchange. For the anion exchange a 1% solution of the Cy5Cl in the selected solvent mixture was passed through the column. After evaporation of the solvent the obtained residue was dried in FV. All steps were carried out at room temperature using gravity as driving force. For the regeneration of the resin a 1 mol/L solution of NaOH was passed through the column until constant pH (12), and then washed with water again until constant pH (5).  C resin = capacity of the resin, n acid = mole before column, n acid unspent = mole after column, m resin = amount of wet anion exchange resin Amberlyst® A26 (OHform).   4)), 7.20 (td, J = 7.5, 0.9 Hz, 2H, H(5)), 7.13 (d, J = 7.9 Hz, 2H, H(17)), 7.08 (d, 7.9 Hz, 2H, H(2)) 6.80 (t, J = 12.5 Hz, 1H, H(13)), 6

Screening of Suitable Solvents on Glass/MoO 3 Substrate
A 20 nm layer of MoO 3 was evaporated on glass to simulate real device conditions. All dyes were spincoated on these glass/MoO 3 substrates from preselected solvents and investigated with AFM.

Determination of Resulting Film Thicknesses After Solution Spincoating
UV-Vis spectra showed different values of absorptivity's in processed films despite that the initial dye concentration was kept constant. Obviously the usage of different solvents produces different films thicknesses. This makes tuning of individual film thicknesses necessary for each dye-solvent system. Ellipsometry was chosen as a measurement technique ( Figure S11, S12). The used substrate is a 1 mm thick microscopic glass coated with a 20 nm MoO 3 layer. The aim is to find the right concentration for each dye that gives 10 nm thick films on MoO 3 layers. For all dyes the chosen spin coating speed was 4000 rpm for 60 sec.

Thermal Behaviour of the Dyes
The samples were weighed under air atmosphere, while the measurement was performed under nitrogen flow in a temperature range from 20-600 °C. From 600-900 °C the sample was exposed to oxygen.

UV-Vis absorbance
To determine the molar extinction coefficient ethanolic solutions of the cyanine compounds of 6.37 x 10 -4 mol/L each were prepared. Subsequently solutions with three different concentrations were prepared by diluting the initial solution (Table S4). All measurements were performed in a 0.1 mm quartz glass cuvette using 99.8 % ethanol as reference for the baseline. The relative molar extinction coefficient for each compound was calculated by dividing the slope of the resulting plots of concentration against absorbance intensity by 10 -1 cm. Figure S4. Concentration dependent absorbance.
The calculated extinction coefficients are summarized in Table S5. Since the cyanine dyes are known for the large exciton binding energy which results in strong bounded electron-hole  First the wavelength was converted into wavenumbers with the following formula: Then the extinction coefficient was calculated for each wavenumber with the following formula:  The peak in Figure S5 was assumed to represent the full band of the lowest energy π-π * transition and was integrated to calculate the oscillator force.

Cyclic Voltammetry for Determination of HOMO/LUMO Energy Levels
Cyclic voltammetry (CV) measurements were performed on a PGStat 30 potentiostat   The Ferrocene solution was prepared qualitatively by adding a spatula of ferrocene in 10 mL of electrolyte solution.
All potentials were referenced to NHE by adopting a potential of +0.72 V vs. NHE for Fc/Fc+ in DMF. [3] The rotating disk was equilibrated before first measurement for 30 min at 3000 rpm. Then the rotation speed was reduced to 50 rpm and was kept constant for all measurements. Before each measurement step the solution was fumigated with argon for 15 min. The solvent window was determined by running 30 cycles from -1.5 V until 1.5 V with a scanning rate of 2 V/s, subsequently the scanning rate was reduced to 0.1 V/s and the baseline curve was recorded. Then the corresponding dye was added to the solution and the above described measurement procedure was repeated. The negative potential window was adjusted to -0.75 V for all chromophores.  The calculations of the E g(el) (HOMO/LUMO gap) are based on the assumption that a positive cathodic current can be referred to a reduction process, while a negative anodic current to an oxidation process. Therefore the oxidation potential corresponds to electron extraction from the HOMO level, while the reduction potential is associated with the electron affinity and indicates the LUMO level. By analysing the recorded spectra graphically it is possible to determine the respective reduction E red(dye) onset or oxidation E ox(dye) onset onset potentials. It is notable that the recorded cyclic voltammograms are showing irreversible processes for all investigated dyes, so that the intersection onset of the corresponding peak has to be chosen as the respective potential. While for the ferrocene which undergoes a reversible process the potential is calculated according to: The potentials were measured against a Ag/AgCl reference. The used conversion constant for ferrocene in DMF is 0.72 V. [3] The correction value against NHE for ferrocene was calculated as follows: Korr. Ferrocene =0.72-1/2( ) To calculate the corrected values for the onset potentials against NHE potential the following assumptions were made: The calculation of the HOMO and LUMO was performed by using empirical equations. [4] The used onset potentials were corrected against NHE as described above.       The orientational polarization or dipole polarisation appears at low frequencies around 10 4 Hz. The n and k values are dependent on the wavelengths. At higher wavelengths however the slope is very low and at a certain wavelength the k value becomes 0.    Figure S13. Cyanine atom numbering.
The cyanines were numbered according to the cif file. In the case of Cy5O 3 SMe and Cy5O 3 SPh compounds with chromophore/anion triplets in the asymmetric unit the chromophores are numbered with additional b/c letters. This deviation from the cif file is necessary for clarity. The relevant anion atoms were numbered according to the cif file. In the case of Cy5O 3 SMe and Cy5O 3 SPh with anion triplets in the asymmetric unit the anions are described with the central S atom as S1/S2/S3.

Statistics of the Cy5TFSI Cell
• Detailed insight reveals a broad deviation in all cell parameters • Indication that the active layer morphology is not optimal Film surface is very smooth, which indicates that the problem lies inside the volume of the active layer The results indicated that residues of TFP could cause broad deviations in cell performance data. Therefore after spincoating of the active layer the device was stored for 16 h at 1 x 10 -6 mbar before further processing. All the average values improve after this additional vacuum treatment this indicates that TFP remains in the thin film volume. For further studies TFP should be avoided, even if it forms a very smooth film surface.   The short circuit current is increased compared to the values obtained after vacuum treatment.

Eff (%)
All other values do not reach the averages obtained after vacuum treatment. Therefore it can be concluded that DIO does not have a significant favorable effect on device parameters.

Active Layer Thickness Variation
A thicker active layer is capable of absorbing more light and generating more free charges.
Most of the organic materials have a very low exciton diffusion length of 10 nm, so that thick light absorbing layers result in charge recombination. To find out if this exciton diffusion length limitation also applies to the synthesized materials four devices with different active layer thickness were prepared. Except for the fill factor which reaches its maximum for an active layer thickness of 5 nm, all the OPV parameters reach their maxima at 10 nm active layer thickness. Therefore it can be assumed that the used compounds have an exciton diffusion length of 10 nm. The thickness dependent trends obtained in external quantum efficiency (EQE) spectra support this assumption.

Aging behavior
Unfortunately, devices all show a fast degradation within a few hours in the dark.
Interestingly this manifests in the relative contributions of the active components to the EQE spectrum. We exemplify this behavior in bilayer solar cells using Cy5O 3 SPh as electron donor and C 60 as acceptor. Figure S20 clearly shows that the cyanine contribution to the EQE decreases more significantly than the C 60 after storing the device for 16 h in a glove box.

DFT Calculations of Cyanine Dyes Based on Single Crystal Data
We have used Density Functional Theory to obtain an insight into the electronic structure and charge distribution for the different anion-chromophore pairs of Cy5O 3 SPh, Cy5TFSI and Cy5O 3 SPhMe and we have considered two different structures. Firstly, we have performed calculations on each inequivalent pair in the asymmetric unit cell, as obtained from the crystallographic data. Secondly, we have further energy-minimised each molecule within the unit cell at the B3LYP/6-31G* level. More specifically, all molecules within the unit cell were brought to their local minimum through an optimization within their local environment using the QM/MM scheme implemented in the NWChem software. All atoms surrounding the molecule of reference, within a radius of 12 Angstrom from the atoms in the QM part, were kept fixed, creating a neutral pocket in which the molecule was then relaxed using B3LYP/6-31G*. Since more than one molecule is present in the unit cell, this procedure was performed in a cyclic fashion, i.e. one molecule was relaxed each time in a self-consistent way until convergence was reached. Following the QM/MM optimization we computed the electronic structure for each anion-chromophore pair. DFT calculations were performed using the Coulomb attenuated CAM-B3LYP exchangecorrelation functional and a split valence double zeta polarised 6-31+G* basis set. Singlet vertical excitation energies have been computed from linear response time dependent density functional theory (TD-DFT). All calculations were performed with the NWChem program, version 6.5 [5] and for the visualization of the orbitals and potential maps we used the MView software. [6] Figure S21 displays the Kohn-Sham HOMO and LUMO orbital plots for the different anionchromophore pairs. We find that for all compounds the HOMO and the LUMO frontier orbitals are localized on the cyanine chromophore. In particular, the LUMO displays a stronger localization over the polymethine chain compared with the HOMO and the delocalization pattern of both levels is not affected by the type of the anion. This is consistent with the cyclic voltammetry measurements suggesting that the anion does not have a strong influence on the HOMO and LUMO energy levels when considering solution or gas-phase conditions. The molecular electrostatic potential maps depicted in Figure S22 can provide information about the charge distribution of each anion-chromophore pair. We observe that the negative charge of the sulfonate anions is mainly localized on the oxygen atoms, while the bistriflylimide anion appears as a large diffuse electron cloud. The general trend is that all chromophores form an electrostatic interaction pocket in the gap between the polymethine chain and the two indolium rings. The sulfonate based anions interact electrostatically with the chromophore within this pocket. This suggests potential nucleophilic attack regions within the chromophore and represents a weak point of cyanine dye salts. Due to the crystallographic environment of the first chromophore-anion pair in Cy5O 3 SPh the anion does not show strong interactions with the chromophore. A weaker electrostatic interaction with the chromophore is also observed for the bistriflylimide anion.