A. Synthesis and characterization

The synthesis of BDT-Cl and the general polymerization procedure are illustrated in Scheme 1. Compound 2 was synthesized via a one-pot Williamson ether synthesis.^{Reference Leclerc, Najari and Zou50} The yield of this reaction is low (25%) due to the dual reactivity of dichlorooctane.^{Reference Griffini, Douglas, Piliego, Holcombe, Turri, Fréchet and Mynar30} Although bromine is more widely used in UV cross-linking, the chlorine-terminated n-octyl sidechain was chosen as the cross-linkable sidechain in this work because the chlorine radical is more reactive than the bromine analogue and improves cross-linking efficiency.^{Reference Mcmurry51} Substituting the cross-linking sidechains on the benzodithiophene unit preserves the fluorine on the thienothiophene unit; preparing a fluorinated thienothiophene monomer with a cross-linkable sidechain is synthetically unfeasible.^{Reference Qian, Xu, Hou, Wang, Hou and Tan52,Reference Chen, Chen and Chen53}

SCHEME 1.

Synthetic procedures of the monomer BDT-Cl and the polymers, PTB-Cls, with different numbers of chlorooctyl sidechains.

By substituting a given number of photocrosslinkable BDT-Cl monomers for BDT-EH monomers of PTB7 in the Stille polycondensation reactions, four cross-linkable PTB polymers, with 12.5, 25, 50, and 100% of BDT-Cl were synthesized.^{Reference Liang, Feng, Wu, Tsai, Li, Ray and Yu48} The polymerizations yielded molecular weights greater than 150 kDa, as measured by GPC (Table I). Due to the high population of linear sidechains, PTB-Cl₁₀₀ obtained in a 1:1 feed ratio of thienothiophene and benzodithiophene exhibited limited solubility in common PV processing solvents, such as CB and DCB. Therefore, the feed ratio of thienothiophene and benzodithiophene was adjusted to 1:1.1 for PTB-Cl₁₀₀ to decrease the molecular weight. The other polymers were prepared from 1:1 thienothiophene:benzodithiophene ratios as these are soluble in CB and DCB due to the solubilizing 2-ethylhexyl groups on the BDT-EH unit.

TABLE I.

Structural characterization and optoelectronic properties of PTB7 and PTB-Cl.

^a Integration ratio determined by ¹H NMR, acquired at 120 °C.

The structures of the polymers were confirmed by NMR spectroscopy. The broad resonance peak at ∼4.4 ppm is assigned to –O–CH₂– protons, and the peak at ∼3.5 ppm is assigned to the –CH₂–Cl group (Figs. S1 and S2). For every 2-ethylhexyl substituent replaced by an n-chlorooctyl group, two –CH₂–Cl protons are introduced and the number of –O–CH₂– protons remains the same. Therefore, the ratio between these two peaks increases from 6:0 for PTB7 to 6:4 for PTB-Cl₁₀₀ (Table I).

B. Optoelectronic properties

The absorption spectra of PTB7 and PTB-Cls in the form of thin films are presented in Fig. 1. All polymers exhibited similar absorption profiles with a strong absorbance at 650 nm, which is typical for PTB derivatives.^{Reference Liang and Yu41} λ_onset of PTB-Cl₁₀₀ was ∼15 nm red-shifted than PTB7, and red-shifts further as the BDT-Cl content increases. As a result, the optical band gap (E _g^opt) decreases from 1.70 eV for PTB7 to 1.66 eV for PTB-Cl₁₀₀ (Table I). A similar trend was observed for other PTB derivatives.^{Reference Liang and Yu41,Reference Liang, Feng, Wu, Tsai, Li, Ray and Yu48} The difference in the sidechain structure likely contributes to the red-shift, as polymers with linear sidechains adopt a closer π–π stacking morphology than those possessing branched sidechains, allowing intermolecular excitations to be activated with lower energy photons.^{Reference Ho, Boudouris and Segalman54} The sidechains on the BDT unit govern the π–π stacking distance of PTB. Therefore, the introduction of linear sidechains results in a closer packing and red-shifted optical absorptions for PTB-Cls. Electrochemical band-gaps of 1.7–1.8 eV were determined by CV (Table I). These results are in good agreement with the reported values for other PTB derivatives.^{Reference Liang and Yu41,Reference Liang, Xu, Xia, Tsai, Wu, Li, Ray and Yu45}

C. Cross-linking studies

Photocrosslinking and insolubilization of cast films were evaluated by measuring the changes of the absorption spectra and the film thickness after UV-irradiation and immersion in the casting solvent, CB (Fig. 2). Thin films cast from their neat CB solutions were subjected to UV exposure followed by an immersion in CB to remove any soluble polymer. All PTB-Cl films were completely insolubilized after 30 min of UV exposure. Those polymers with high contents of cross-linkable groups (i.e., PTB-Cl₁₀₀, PTB-Cl₅₀) required only 10 min exposure. By contrast, PTB7 remained soluble, even after long UV exposures. With insufficient UV exposure, PTB-Cl films peeled away from the ITO substrates and broke into insoluble pieces upon immersion into CB. The cross-linking process is represented here by the films of PTB-Cl_12.5: as UV exposure increased from 10 to 30 min, the detached, insoluble film fragments became larger with increasing irradiation dose and ultimately the film remained attached to the substrate, intact after 30 min irradiation (Fig. S5). This suggests that photocrosslinking was homogeneously initiated throughout the film and propagates laterally and vertically until a cross-linked network is formed. The surface temperature of the UV irradiation platform increased to only 26 °C (slightly above room temperature) after 30 min of continuous UV irradiation, indicating that cross-linking was photochemically activated, not thermally activated.

FIG. 2.

Photocrosslinking of PTB-Cls evaluated using (a) UV-vis spectroscopy and (b) profilometry of the remaining film after irradiation and immersion into solvent.

To demonstrate photoinduced cross-linking, photolithography was performed using PTB-Cl films. The UV exposure and solvent immersion processes were similar to cross-linking studies except masks (TEM copper grids, bar width: 35 μm) were placed on these polymer films to create patterns. Optical microscopy images of photolithographed films reveal that clear images of the copper grid were obtained with the PTB-Cl polymers, with PTB-Cl₁₀₀ displaying the sharpest edges (Fig. 3). These preliminary results demonstrate the potential to fabricate high-performance microcircuits using cross-linked PTB derivatives.

FIG. 3.

Optical microscopy images of patterned PTB-Cl₁₀₀ (a), PTB-Cl₅₀ (b), and PTB-Cl₂₅ (c). Scale bar: 200 μm.

The mechanism of photocrosslinking of PTB-Cl is believed to be free radical in nature. Similar hypotheses have been proposed in photocrosslinking studies of conjugated polymers bearing bromine-functionalized sidechains.^{Reference Kim, Miyamoto, Ma and Fréchet29,Reference Yau, Wang, Treat, Fei, Tremolet de Villers, Chabinyc and Heeney33} The photochemical process likely involves a series of chain reactions that begin with UV-initiated chlorine-free-radical formation.^{Reference MacKenzie and Harrison55} The proposed mechanism is illustrated in Scheme 2. Upon UV irradiation, homolysis of the C–Cl bond generates a chlorine radical, leaving a radical species on the end of the sidechain. The highly reactive chlorine radical may randomly attack another alkyl chain by abstracting a hydrogen from the C–H bond. The radical species on the alkyl chains can terminate by forming new carbon–carbon bonds, which constitutes cross-linking.

SCHEME 2.

Possible mechanism of photocrosslinking for PTB-Cl: (a) homolytic photolysis of the C–Cl bond; (b) hydrogen abstraction; (c) and (d) termination of two radical species.

XPS was performed to verify the hypothesized mechanism. PTB-Cl films were examined before and after 30 min of UV exposure. UV exposure was performed while the samples were affixed to the XPS sample holder, and measurements were carried out on the precise same regions before and after UV exposure. XPS revealed a decrease of the chlorine peaks after UV exposure while other peaks remained unchanged (Fig. S6). The loss in chlorine is most prominent in the PTB-Cl₁₀₀ sample, where 20% was lost after 30 min of UV irradiation.

D. Photovoltaic properties

Bulk heterojunction photovoltaic devices of PTB7 and PTB-Cls were fabricated in a conventional ITO/PEDOT:PSS/polymer:PC₇₁BM/Ca/Al configuration and tested under standard 1 sun AM 1.5G illumination. The active layers were spin-coated from a DCB solution of polymer:PC₇₁BM with a thickness of ∼110 nm. J–V curves and EQE curves are presented in Fig. 4, and PV parameters are tabulated in Table II. PTB-Cl devices showed similar PCEs to PTB7 devices at around 4.0–4.5% with similar FF, V _oc, and J _sc values. This result indicates that the introduction of photocrosslinkable sidechains on the BDT unit did not significantly alter the PV performance of the polymers.

FIG. 4.

J–V curves (a) and EQE curves (b) of PTB7 and PTB-Cl devices cast from DCB.

TABLE II.

Photovoltaic characteristics of PTB7 and PTB-Cl devices.

The influence of photocrosslinking on the PV performance was investigated. For all the PTB-Cl polymers, 10-min UV exposure had a significant impact on the J–V characteristics. Both J _sc and V _oc values fell dramatically after UV irradiation (Fig. 5). This effect is most severe in PTB-Cl₁₀₀ devices, for which photocrosslinking is the heaviest. Interestingly, J _sc and V _oc increased slightly after thermal annealing following the UV exposure. The same effect has been observed in other photocrosslinkable polymers with bromine-terminated sidechains.^{Reference Griffini, Douglas, Piliego, Holcombe, Turri, Fréchet and Mynar30,Reference Yau, Wang, Treat, Fei, Tremolet de Villers, Chabinyc and Heeney33} Indeed, conjugated polymers can be reversibly p-doped with oxygen radical anions, forming a charge-transfer complex (CTC),^{Reference Abdou, Orfino, Son and Holdcroft56,Reference Desiraju and Steiner57} which induces trapping sites and causes misalignment of energy levels between the polymer and the fullerene, leading to losses of J _sc and V _oc.^{Reference Schafferhans, Baumann, Deibel and Dyakonov58–Reference Aygu, Hintz, Egelhaaf, Distler, Abb, Peisert and Chassé60} The formation process of CTC is accelerated by illumination while the dissociation process is activated by thermal annealing.^{Reference Abdou, Orfino, Son and Holdcroft56,Reference Seemann, Sauermann, Lungenschmied, Armbruster, Bauer, Egelhaaf and Hauch59,Reference Liao, Yang, Liu, Horng, Meng and Shy61–Reference Aguirre, Meskers, Janssen and Egelhaaf64} Therefore, we attribute the UV degradation and thermal recovery of performance to reversible p-type doping of the polymer by chlorine radicals. To compare, a PTB7 device was also subjected to 10 min of UV irradiation followed by thermal annealing. Unlike the PTB-Cls, the UV degraded J _sc and V _oc were not recovered upon thermal annealing. Therefore, the partially recoverable UV degradation of PTB-Cl devices was a combined effect of the reversible p-doping and certain permanent degradations.

FIG. 5.

J–V curves of (a) PTB-Cl₅₀ and (b) PTB7 devices: as-cast (solid), after 10-min UV exposure (dotted), and subsequently thermally annealed (dashed).

E. Stability of PV devices

Evaluation of stability of PV devices was carried out by accelerated aging at 120 °C inside the glove box and measuring the PV parameters at the end of 1, 3, 5, and 10 h of heat treatment. To find an optimal photocrosslinking condition that introduces sufficient cross-linking to stabilize the cell performance without a negative impact on the PV parameters, UV exposure times of less than 10 min (e.g., 30 s, 1 min, 2 min, and 5 min) were investigated. PCE profiles plotted against the time of thermal treatment are shown in Fig. 6. The PCE of a PTB7 device degraded from 4.01 to 2.13% after 10 h at 120 °C. PTB-Cl₁₀₀, PTB-Cl₅₀, and PTB-Cl₂₅ devices with 30 s, 1 min, and 2 min UV exposures exhibited more stable PCEs. The UV degradation became significant as the UV exposure was extended beyond 5 min, leading to poor PCEs. PTB-Cl₅₀ devices with short UV exposures exhibited the highest stable PCEs among all devices. The slight increase of PCE of the 1 and 2 min UV-exposed PTB-Cl₅₀ devices after 1 h at 120 °C is associated with the thermal recovery of UV degradation. With short UV exposure, the UV degradation is fully recoverable because no permanent damage is caused. The high stability of the UV irradiated samples is a combined result of stable J _sc and FF (Figs. S7 and S9). By contrast, non-crosslinked PTB-Cl films showed poor thermal stability, caused by a rapidly decreasing FF.

FIG. 6.

PCE as a function of thermal treatment time at 120 °C of PTB-Cl₁₀₀/PC₇₁BM (a), PTB-Cl₅₀/PC₇₁BM (b), PTB-Cl₂₅/PC₇₁BM (c), and PTB-Cl_12.5/PC₇₁BM (d) devices with various amounts of UV exposure, compared to the PTB7/PC₇₁BM device.

F. Morphological stability

Optical microscopy was performed on solar cell devices to investigate the correlation between the evolution of morphology and the degradation of PCE. No PCBM crystallization was observed on any of the devices thermally treated at 120 °C for 10 h. Hence, the degradation of PV parameters observed is not caused by the crystallization of PCBM, as determined by optical microscopy. However, nanoscale phase segregation that is not in the form of PCBM crystallization has been found to result in degradation of PV devices.^{Reference Schaffer, Palumbiny, Niedermeier, Jendrzejewski, Santoro, Roth and Müller-Buschbaum65}

To examine whether cross-linking inhibits gross, microsized phase segregation—the end result of nanoscale phase segregation—optical microscopy was performed on cross-linked films exposed at a higher temperature, 150 °C, for up to 20 h. The “active layer” morphologies of the polymer/PC₇₁BM devices after 20-h thermal treatment are presented in Fig. 7. For pristine PTB-Cl devices, a large density of PCBM crystallites is observed after 20 h at 150 °C, which increases as the BDT-Cl content increases. However, only 10-min UV exposure was sufficient to completely inhibit the crystallization of PCBM in all PTB-Cl devices. Shorter UV exposure times are more effective as the BDT-Cl content decreased. In PTB7/PC₇₁BM devices, only sporadic crystallites were found after 10 h at 150 °C (Fig. S10). Extending the heating to 20 h produced large crystallites with dimensions of 10–20 μm in length. The blue halos surrounding the crystallites represent regions depleted of PCBM.^{Reference Zhong, Yang, deWith and Loos66,Reference He, Germack, Joseph Kline, Delongchamp, Fischer, Snyder, Toney, Kushmerick and Richter67}

FIG. 7.

Optical microscopy images of the active layers of pristine and UV exposed PTB-Cl devices and PTB7 device after 20 h at 150 °C. Scale bar: 20 μm.

The differences in crystallite size, number, and onset time for crystallite formation are noticeable between PTB7 and PTB-Cl₁₀₀ devices. Densely populated small crystallites (∼2 μm) were found in PTB-Cl₁₀₀ devices. This observation indicates that the nucleation rate is greater than the growth rate, such that the PCBM molecules tend to form new nuclei rather than adding to existing crystallites. By contrast, the crystallization of PCBM in the PTB7 device is nucleation limited, i.e., the growth rate is higher than the nucleation rate, resulting in large, sparsely populated crystallites. We attribute the difference in the PCBM crystallization kinetics of PTB7 and PTB-Cl₁₀₀ films to the structural dissimilarity of the sidechains on the BDT unit. PTB-Cl₁₀₀ with linear chlorooctyl sidechains permits tighter π–π stacking than PTB7 with its branched 2-ethylhexyl sidechains. The closely packed PTB-Cl₁₀₀ expels the PCBM molecules from the mixed phase to form nuclei, whereas the loosely packed PTB7 hinders this process.

While light-density cross-linking in PTB-Cl_12.5 is sufficient to retard nucleation, as evidenced by the absence of PCBM crystallites after 20 h at 150 °C, this was not effective to counterbalance the increased nucleation rate as the BDT-Cl content was increased. Noticeably, no PCBM crystallite was observed near the top electrode for all samples after the 20-h thermal treatment (Fig. S11), suggesting that the electrode inhibits crystallization of PCBM, which corroborates with the confinement effect of the top electrode.^{Reference Yang, Alexeev, Michels and Loos68,Reference Watts, Belcher, Thomsen, Ade and Dastoor69}