DOI: 10.1002/ ((please add manuscript number))
Article type: Full Paper
Investigation of Radical and Cationic Crosslinking in High Efficiency, Low Band Gap Solar cell Polymers.
By Chin Pang Yau†, Sarah Wang†, Neil D. Treat, Zhuping Fei, Bertrand J. Tremolet de Villers, Michael L. Chabinyc*, Martin Heeney*
Dr Chin Pang Yau, Dr Zhuping Fei and Prof. Martin Heeney
Department of Chemistry and Centre for Plastic Electronics
Imperial College London
Exhibition Road, South Kensington
London SW7 2AY
UK
E-mail:
Sarah Wang, Dr Neil D. Treat, Dr Bertrand J. Tremolet de Villers and Prof Michael L. Chabinyc
Materials Department
Elings Hall Room 3219
University of California
Santa Barbara CA 93106-5050
U. S. A.
E-mail:
†Both authors contributed equally.
Keywords: Dithienogermole, Organic Solar Cell, conjugated polymer, Crosslinking
Abstract
We report the synthesis and characterization of dithienogermole-co-thieno[3,4-c]pyrroledione (DTG-TPD) polymers incorporating chemically crosslinkable sidechains and compare their properties to a parent polymer with simple octyl sidechains. Two crosslinking groups and mechanisms are investigated, UV promoted radical crosslinking of an alkyl bromide crosslinker, and acid promoted cationic crosslinking of an oxetane crosslinker. We find that random co-polymers with a 20% incorporation of the crosslinker demonstrate a higher performance in bulk heterojunction solar cells than the parent polymer, whilst 100% crosslinker incorporation results in deterioration in device efficiency. The use of 1,8-diiodooctane (DIO) as a processing additive improved as-cast solar cell performance, but was found to have a significant deleterious performance on solar cell performance after UV exposure. The instability to UV could be overcome by the use of an alternative additive, 1-chloronapthalene, which also promoted high device efficiency. Crosslinking of the polymer was investigated in the presence and absence of fullerene highlighting significant differences in behavior. Intractable films could not be obtained by radical crosslinking in the presence of fullerene, whereas cationic crosslinking was successful.
1. Introduction
The surge of interest in organic photovoltaics over the last few years has resulted in a steady increase in device efficiencies, with values now approaching thresholds for commercialisation.1 Organic devices are typically comprised of a mixture of two components in a bulk heterojunction (BHJ), with the device performance crucially dependent on the complex nanostructure of this blend. This nanostructure is influenced by the miscibility of individual components, as well as their tendency to crystallise and phase separate.2 Desirable morphologies are thought to contain bicontinuous domains with sizes on the order of 10 nm for effective charge generation and separation.2-3 This morphology has been typically achieved by the manipulation of coating conditions, post processing thermal or solvent annealing and by the choice of solvent and processing additives.2-3 In many cases the nanostructure of the blend can be considered kinetically trapped rather than thermodynamically preferred.4 As such, the as-cast morphologies can change over time, because the fullerene in polymer:fullerene BHJs is able to diffuse within the amorphous regions of the polymer matrix.5 Therefore the lifetime of the device may be compromised by the reorganisation of fullerene and polymer to a more thermodynamically stable state over time. This change can be particularly relevant during device operation, in which relatively high temperatures can be reached. One possible approach to overcome this issue is to ‘lock-in’ the optimum kinetically trapped morphology by some form of chemical crosslinking after the active layer has been formed.
Crosslinking of donor polymers has been shown to stabilise the nanostructure in polymer:fullerene BHJs, principally by restricting the ability of PCBM, the most common fullerene acceptor, to crystallise or aggregate into large domains in thin films.6 Crosslinkable polymers also have other potential benefits, such as their use in bilayer devices where deposition of the second layer can be facilitated by crosslinking of the first. For example some element of structural control over phase connectivity and length scale has been recently achieved by diffusion of fullerene derivatives into lightly cross-linked donor polymers during solution deposition.7 In these examples swelling of the polymer network by the solvent assisted the diffusion of fullerene into the film. Therefore, in general, polymer crosslinking may offer more flexibility in methods to form BHJs, as well as providing methods to control and stabilise the nanostructure in BHJs.
Crosslinking groups have been incorporated onto the lateral sidechains of conjugated polymers to enable either thermal or photochemical crosslinking.6b Reactive groups such as alkyl bromide, oxetane, azide and alkenyl have been reported as effective crosslinkers. Many of these studies have used modified regioregular P3HT to demonstrate the proof of concept. 6a,8 However, there are relatively few examples in which low band gap donor-acceptor polymers, which provide higher power conversion efficiency, have been investigated.6c,9 Within the small class of low band gap polymers reported thus far, device efficiencies have been rather modest even before crosslinking. Furthermore there have been few studies examining the differences in performance for radical and cationic crosslinking groups.
Two different reactive crosslinkers have previously demonstrated promising performance in semiconducting polymers, alkyl bromide groups as a radical crosslinkers and oxetanes as a cationic crosslinkers. Alkyl bromide functionalised sidechains have been demonstrated as a promising route to stabilise blend morphology in both P3HT functionalised polymers, and in low band gap benzodithiophene copolymers.6c,8c The process of crosslinking is believed to be a radical based mechanism, whereby direct exposure to 254 nm UV light causes homolytic cleavage of the carbon-bromine covalent bonds.10 Subsequent crosslinking likely occurs by abstraction of a proton by the bromine radical from the sidechains of adjacent polymers, followed by coupling of the alkyl radicals. The crosslinking of oxetane groups has been widely investigated in OLED materials.11 Oxetane groups polymerise via a cationic ring opening mechanism also known as CROP.12 In comparing oxetane to other cationic crosslinking groups like epoxides, the oxetane functionality has the advantage of high stability to a variety of chemical conditions, facilitating the synthesis of oxetane containing materials. Furthermore unlike radical based crosslinking, which is sensitive to oxygen, cationic polymerisation can be performed in ambient atmosphere. In addition the cationic reactive center is generally less reactive and less prone to side reactions than highly reactive and non-selective radicals. The CROP mechanism typically requires small amounts of a photo-acid generator (PAG) to initiate the reaction. Following initiation, heat is generally applied in the form of a post-exposure bake in order to accelerate the crosslinking. Although several conjugated polymers incorporating oxetane functionality have been successfully crosslinked and photo-polymerised for OLED applications, at the outset of this work there were few examples in OPV.8a,13
In this paper we report the synthesis and characterisation of random alternating donor-acceptor co-polymers incorporating either oxetane or alkyl bromide crosslinkers, and compare their optoelectronic properties to the parent alternating polymer. We focus on thieno[3,4-c]pyrrole-4,6(5H)-dione (TPD) as a suitable readily functionalised electron poor co-monomer. TPD has been utilised in various high performing donor-acceptor co-polymers, and generally promotes high open circuit voltage and good photocurrent. 14 As an electron rich co-monomer, we investigated dithienogermole (DTG) due to the promising performance of the PDTG-TPD co-polymer previously reported, and our experience with this comonomer.15 We find that the incorporation of small amounts (20%) of crosslinkable monomer has a beneficial impact on overall device efficiency before crosslinking. The role of various processing additives is investigated, and we find the that the use of 1,8-diiodoctane (DIO) causes unexpected stability issues when the films are irradiated with UV radiation. The use of chloronapthalene (CN) as an alternative additive circumvents these issues. Finally we compare and contrast radical and cationic crosslinking in blends with fullerene derivatives and demonstrate that the presence of fullerene in the blend reduces the efficacy of radical crosslinking.
2. Results and Discussion
2.1. Synthesis
4,4-Bis(2-ethylhexyl)-2,6-bis(trimethylstannyl)-4H-germolo[3,2-b:4,5-b']dithiophene16 DTG-Sn2) and 1,3-dibromo-4H-thieno[3,4-c]pyrrole-4,6(5H)-dione6c (TPD) were synthesised by literature procedures. However, the purification of DTG-Sn2 was complicated due to its chemical instability towards column chromatography over silica. Therefore, purification was performed by preparative gel permeation chromatography (GPC) on a crosslinked polystyrene size exclusion column using hexane as the eluent. The TPD oxetane, bromide and octyl variants were synthesised according to a modification of literature procedure6c, as shown in Scheme 1a. Hence TPD was deprotonated with sodium hydride in DMF, and the appropriate alkylating agent was subsequently added. In the case of 1,6-dibromohexane, excess reagent was added in order to prevent the reaction of two TPDs with one 1,6-dibromohexane.
Alternating co-polymers incorporating 100% of the crosslinking groups, as well as random co-polymers incorporating 20% of the crosslinker and 80% of an octyl substituted TPD co-monomer (TPD-C8) were prepared, along with the parent PDTG-TPD-C8 co-polymer. Scheme 1b shows the nomenclature used to designate the percentage crosslinker in the polymer backbone. Percentages refer to the statistical amounts of co-monomers added at the beginning of the reaction, thus the 20% statistical co-polymers were synthesised using a 0.2 mol equivalence of the desired crosslinking TPD monomer (TPD-Ox or TPD-Br) with 0.8 mol equivalence of the TPD-C8. Five co-polymers, shown in Scheme 1c, were synthesised via a microwave assisted Stille coupling method in chlorobenzene.17 After precipitation, they were dissolved in chloroform then washed in sodium diethyldithiocarbamate trihydrate to extract any residual palladium before being precipitated into methanol.18 Soxhlet extraction was performed with methanol, acetone and hexane to remove catalyst residues and low weight oligomers. Finally, the remaining polymer was extracted into chloroform and the polymer was further fractionated by preparative GPC in hot chlorobenzene (80 °C). This method was used to modulate the molecular weights and narrow the polydispersity (PDI) for all five PDTG-TPD polymers.
All polymers demonstrated reasonable solubility in common organic solvents like chloroform and chlorobenzene upon warming. The molecular weights were measured by GPC relative to polystyrene standards, as shown in Table 1, which shows the number average molecular weight (Mn), degree of polymerisation for Mn (DPn), weight average molecular weight (Mw), degree of polymerisation for Mn (DPw) and polydispersity (PD) of all five polymers. The molecular weight of PDTG-TPD-C8 was similar to that previously reported by Reynolds et al.,15b and both of the random co-polymers showed comparable molecular weights, with degrees of polymerisation around 60-70 for all polymers. For the PDTG-TPD-Ox100% and PDTG-TPD-Br100% copolymers, slightly lower Mn‘s were observed, with degrees of polymerisation around 35.
The composition of the random co-polymers was confirmed by 1H NMR in CDCl3 (see Supporting Information). Although the aromatic protons and methylene groups close to the polymer backbone were rather broad and poorly resolved, likely due to restricted rotation, the peaks attributable to the alkyl bromide or oxetane were sharp. Integration of the aromatic protons (8.60-7.30 ppm) and the oxetane proton environment (4.51 and 4.35 ppm) gave a ratio of 1:0.38, within the experimental error of the expected value of 1:0.4. Similarly, for PDTG-TPD-Br20% the ratio was observed to be 1:1.1 (aromatic DTG proton environment (8.60-7.30 ppm):CH2 adjacent to the TPD nitrogen + CH2 adjacent to the bromine on TPD-Br (3.71-3.43 ppm)) which is in agreement with the theoretical reaction stoichiometry.
Differential scanning calorimetry (DSC) and thermogravimetric analysis (TGA) were performed under an inert atmosphere of nitrogen on all co-polymers. DSC measurements showed no obvious transitions from -20 °C to 330 °C for all co-polymers. TGA of PDTG-TPD-C8, PDTG-TPD-Ox100% and PDTG-TPD-Br20% shows 5% mass loss at 431 °C to 433 °C, while PDTG-TPD-Ox20% shows a 5% mass loss at 450 °C and PDTG-TPD-Br100% at 385 °C (Figure S6). The TGA shows that there are no low temperature decompositions, and since the chemical composition of the backbone is similar for all polymers, it suggests that the 100% alkyl bromide incorporation may be the reason for the earlier decomposition temperature of PDTG-TPD-Br100%. The temperatures where decomposition occurs are much higher than any annealing temperatures used in the fabrication process for BHJs making these polymers highly stable for use in solar cells.
2.2. Optical and Electrochemical Properties
The UV-Vis spectra were measured in chlorobenzene solution at room temperature and as thin films spin coated from a solution of chlorobenzene (5 mg/mL) (Figure 1). The long wavelength onset for all polymers is 730 nm, corresponding to an optical band gap (Eg) of 1.7 eV. All polymers exhibited similar shape spectra for both solution and film measurements, with a pronounced series of peaks with the most intense around 680 nm and 620 nm (Table 1), similar to that previously reported.15b This separation (~0.2 eV) is consistent with a vibronic progression as observed in many ordered semiconducting polymers. There was almost no change in the position of the absorption maxima (±2 nm shifts) upon film formation suggesting aggregation in solution. Upon comparison of the five polymers, it is clear that the parent alkylated polymer (TPD-C8) and both polymers with 20% crosslinker groups show almost identical spectra, with very little difference in the relative heights of the two absorption peaks, both in solution and thin film. However, for the 100% oxetane and bromine containing polymers it can be seen there is a change in the relative peak amplitude of the two maxima, with the long wavelength peak reducing in relative absorption versus the shorter wavelength peak. Since the relative ratios of these two peaks are often related to coupling of the transition dipoles in the solid state, this suggests that high amounts of the relatively large crosslinking groups are influencing the intermolecular packing of the conjugated backbones. We do not believe this is related to the lower degree of polymerisation for both of the 100% polymers, because the MW should be sufficient to be above the effective conjugation length in both cases. Additionally, PCPTD-TPD-C8 of much lower weight (Mn 16.3 KDa) has been reported to have an identical UV spectra to that of the higher weight polymer.15e Therefore, we believe the origin is more likely related to the high percentage of crosslinking groups changing the interactions of the sidechains.
The ionisation potential of polymer thin films was determined using photo-electron spectroscopy in air (PESA) and is shown in Table 1. The reduction potential (LUMO) was estimated by the addition of the optical Eg to the ionization potential. The ionization potentials are all in the range of 5.2 to 5.3 eV, and fall within the standard error of 0.05 eV associated with this particular measurement. These results support the fact that the 20% and non crosslinkable polymer are optically and electronically very similar. For both of the 100% crosslinkers (Br and Ox), a slightly lower HOMO of 5.3 eV was observed, which could be due to the crosslinker functionality disrupting the packing and backbone planarisation.